best available copy for all picturessigns and experimental results for coatings for alkali halides,...

Best Available

Copy for all Pictures

: .-■■. ..-,-:>:v.-, I ■■jtMii)»)WrM.'ftffy»JW B

Z'

V.

AD-766 479

ADVANCED MODE CONTROL AND HIGH-POWER OPTICS

TECHNOLOGY

VOLUME II. HALIDE WINDOW MATERIALS TECHNOLOGY

HUGHES RESEARCH LABORATORIES

PREPARED FOR

AIR FORCE WEAPONS LABORATORY

JULY 1973 DISTRIBUTED BY:

mr National Technical Information Service U. S. DEPARTMENT OF COMMERCE

- ■ ■-- ■ — ^■-* -^ ^..-- -- -^riiniMiriiiiiiiiiiii-iiitiiiiiiili'iiiiiiiiiiiirfiiiaiiiiii mi -- ^..^-^^w^^^.,.,^, .^■■. ~t..-^.*.^..Ä.±-l^^..J ^MM

—~™»B»«W " " twmmmimm -m^~m~m^*~*l IWmJIIMW.nmWl'WWPPWlPBWII^PIWPBiWIlPI'l««"""»^^ Trnvwemmmm

CbuJ 6

AFWL-TR-72-152, Vol II AFWL-TR- 72-152, Vol II

A

CO

A

Ü 1/

k*

ADVANCED MODE CONTROL AND

HIGH-POWER OPTICS TECHNOLOGY

Volume II

Halide Window Materials Technology

R. C. Pastor

M. Braunstein

Hughes Research Laboratories

TECHNICAL REPORT NO. AFWL-TR-72-152, Vol II

■ - ''

July 1973 SEP 18 1374

... ■. i

■

B

AIR FORCE WEAPONS LABORATORY

Air Force Systems Command

Kirtland Air Force Base

New Mexico leproducBd by

Spn-ngHeld, VA. «IS*'"

Approved for public release; distribution unlimited.

6X

i IMUMIBW^"'1-!^^ ..M.VK..I ûî^^^p^îpx^nîn^^pwiwipqpî^ m.w»ifNiiwaJW«l*.L^<i|Hi tijjjuiwiwij.i .iWjiim.>pjw^«»«iiHwiswuww.wiipiM^JWW^wm^pi(wwiy..^

AFWL-TR-72-152, Vol II

HCCESSION (ar

•ms

■ ^

■

White Sfitlcn

D

DY . STRUBTIO« AVAlLABILin CODES

Disl. A>*IL. Mil or SFüCIAL

fl

AIR FORCE WEAPONS LABORATORY Air Force Systems Command Kirtland Air Force Base

New Mexico 87117

I

When US Government drawings, specifications, or other data are used for any purpose other than a definitely related Government procurement operation, the Government thereby incurs no responsibility nor any obligation whatsoever, and the fact that the Government may have formulated, furnished, or in any way supplied the said drawings, specifications, or other data, is not to be regarded by implication or otherwise, as in any manner licensing the holder or any other person or corporation, or conveying any rights or permission to manufacture, use, or sell any patented invention that may in any way be related thereto. .

DO NOT RETURN THIS COPY. RETAIN OR DESTROY.

■ -•MJl

'-1 - - ■ ' — ^ ■ - ■ - ^^^^^ Müku

r-Tir-i'—7--—i . . ^1.^1. iv«T»i-«m,«.'*^«!^j r^rr ■ ■■ ■ .,-l,.^P,--m "i-^r- — -rr- —-■- .—.„,m,,r

UNCLASSIFIED _SeriMTljv Classification

DOCUMENT CONTROL DATA -R&D (Srcunty ctmssltleatlon of tl'le, body ol abslrecl and indexine annolalion nm.sl he entered when the overall report Is clessllled)

ORIGIN* TING ACTIVITY (Corporate author)

Hughes Research Laboratories 3011 Malibu Canyon Road Malibu, California 90265

3 REPORT Tl TLE

2a. REPORT SECURITY CLASSIFICATION

UNCLASSIFIED 2b. GROUP

ADVANCED MODE CONTROL AND HIGH-POWER OPTICS TECHNOLOGY, Vol II, Halide Window Materials Technology

4 DESCRIPTIVE NOTES (Type ol report and Inclusive dates)

18 October 1971-31 October 1972 5 AUTHORISI (First name, middle initial, last name)

R. C. Pastor; M. Braunstein

« REPORT DATE

July 1973 »«. CONTRACT OR GRANT NO

b. PROJEC T NO.

cTask Nos 0 and 8

F29601-71-C-0101

1256 and 317J

70. TOTAL NO. OF PAGES

150 76. NO. OF REFS

i 30 9a. ORIGINATOR'S REPORT NUMBERIS)


10 DISTRIBUTION STATEMENT

lb. OTHER REPORT NO(S) (Any other numbers that may be assigned this report)

Approved for public release; distribution unlimited,

II SUPPLEMENTARY NOTES

13. AOSTRAC T

12. SPONSORING MILITARY ACTIVITY

AFWL (LR0) Kirtland AFB, NM 87117

(Distribution Limitation Statement A) " " The results reported here represent significant advances in the state of the art of high-power laser windows and coatings. An order-of-magnitude improvement was achieved in the material parameters of single crystal KC1 over the values which characterized the best available materials at the beginning of the study. Single crystal specimens of KC1 (up to 10-cm diameter) having optical absorption at 10.6 ym <^ 0.0005 cm"1 and rupture strengths as high as 4000 psi were produced using a reactive atmosphere processing (RAP) technology which was developed during the program. The RAP technique, involving processing under an atmosphere of He + CCU, was used for window materials purification, crystal growth, and for surface layer exchange in KC1 crystals. It led to the development of a one-step processing for KC1 crystal growth, from powder to crystal, which minimizes the incorporation of OH" in the KC1 crystals. Optical, mechanical, and surface stability evaluations of various fluoride and chloride crystals are also reported. The coating program goal, which was to achieve 0.1% or less absorption per coated window surface at 10.6 ym, has been demonstrated for some of the optical coatings prepared during the program. Theoretical designs and experimental results for coatings for alkali halides, zinc selenide, and cadmium telluride are presented with emphasis placed on coatings for KC1. Rigorous control of starting material purity and substrate preparation was found to be essential for the preparation of low-absorption coatings.

DD FORM 1 NOV 85 1473 UNCLASSIFIED

Security Classification

--"- -g-^^. _.^.—._>—^—L_M_^-i^».^^-J—^— ^-. ._.. . . ....

•Kmmrm^nnwi ii n i u wmw ■ i wratoiwaii i i i ••!■ win *^m-*r**m^m'm***mmmn*vm npi juiia i ui mmtin ■■ i n aB^vwnBwswm — I" i i-i"»™»«*!

UNCLASSIFIED Security Classification

K EV WORDS

Laser window materials Laser window coatings Antireflection coatings Protective coatings Halide materials

LINK C

jOs mi &SSIEIED

Security Classification

- -■-■■-- - - — •■^. ■—->-■..■■ . - ,.: ^.^■.■,:...-. ^,. -..,.,. ■ ,.

«Twpnf^n


ADVANCED MODE CONTROL AND HIGH-POWER OPTICS TECHNOLOGY

I Volume II

R. C. Pastor

M. Braunstein

Hughes Research Laboratories

TECHNICAL REPORT NO. AFV1L-TR-72-152, Vol II

Approved for public release; distribution unlimited,

lh

— — - ■ ——- -. -

^Mnm»mMpn>mT^^T^-—i .^ .> ■ uJ.-^.^"■ ■■ ■ > ».'■-.<^jnM^wfM^nifpBPinpppiW^.ffWMW^ '^M«^-wwPMWwwii^«w*ffi^"»î'w^.m"«.wi. wwi.-fflP'^^.i-«n-p-««!""-,ip«rT"'i-iB ■ ^ u JW .«M-H. PM-f. I, wi^P-iPM-M > ■■»*■ i -■»■ «u i . IU jiimm

AFV;L-TR-72-152, Vol II

FOREWORD

This report was prepared by Hughes Research Laboratories, Malibu, California, under Contract F29601-71-C-0101. The research was performed under Program Elements 611010 and 63605F, Projects 1256 and 3170, Tasks 0 and 8. The research was partially funded by the Advanced Research Projects Agency under ARPA Order Number 1256.

Inclusive dates of research were 18 October 1971 through 31 October 1972. The report was submitted 18 May 1973 by the Air Force Weapons Laboratory Project Officer, Captain Harry V. Winsor (LRO).

The following personnel participated in various phases of the study: A C. Pastor; M. Robinson; K. Arita; A. Timper; R. Turk; K. Robinson; A. I. Braunstein; J. E. Rudisill; D. H. Williams; C. E. Pilon; J. E. Kiefer; and J. H. Yoder.

This technical report has been reviewed and is approved.

^M Ü /U—^-O-V. HARRY V. WINSOR Captain, USAF Project Officer

EDWARD N. LAUÖHU^/ A Lt Colonel, ÖSAF & Chief, Optics Technology Branch

x. RUSSELL K. PARSONS Colonel, USAF Chief, Laser Division

ii

1*~.*..^...:... ^-.-, -■ - —U - - - -- üii^MiÄjtÛjî-^,-^,

1..W ...(«, M1l.i...J.J».p™lli|1_ I .^HL,^

ABSTRACT

(Distribution Limitation Statement A)

The results reported here represent significant advances in the state of the art of high-power laser windows and coatings. An order- of-magnitude Improvement was achieved in the material parameters of single crystal KC1 over the values .vhich characterized the best available materials at the beginning of the study. Single crystal specimens of KC1 (up to 10-cm diameter) having optical absorption at 10.6 pm < 0.0005 cm-1 and rupture strengths as high as 4000 psi were produced using a reactive atmosphere processing (RAP) technology which was developed during the program. The RAP technique, involving processing under an atmosphere of He + CCl., was used for window materials purification, crystal growth, and for surface layer exchange in KC1 crystals. It led to the development of a one-step processing for KC1 crystal growth, from powder to crystal, which minimizes the incorporation of OH" in the KC1 crystals. Optical, mechanical, and surface stability evaluations of various fluoride and chloride crystals are also reported. The coating program goal, which was to achieve 0.1% or less absorption per coati I window surface at 10.6 pm, has been demonstrated for some of the optical coatings prepared during the program. Theoretical designs and experimental results for coatings for alkali halides, zinc selenide, and cadmium telluride are presented with emphasis placed on coatings for KC 1. Rigorous control of starting material purity and substrate preparation was found to be essential for the preparation of low- absorption coatings.

iii/iv

liTh.«!«!.» rl■r|•^ill|•MHlfr•^^l(th■|■|^lllr■ill■vfl■^l^■ll^l^llfl■^Tr<^^^^^'^^^^a"^"^-^^^•^^^"^'- -^---"•^--^■--■--- - --■'■ --■■- -■-

'?^^!*P^W!WWW»??Tr^",rwww»7™"W»w*^^

TABLE OF CONTENTS

Secti on

I

II

III

Page

INTRODUCTION AND SUMMARY 1

RESULTS AND EVALUATION 10

A. Window Materials 10

B. .•lindow Coatings 72

CONCLUSIONS AND RECOMMENDATIONS 98

A. i'li ndow Materi al€ 98

B. Window Coatings .' . . 101

APPEND!* I - Reactive Atmosohere Process- ing of Al kal i H al i des 103

APPENDIX II - Optical Absorption of Potassi urn Ch lo ri de at Ten Mi crons 109

APPENDIX III - Absorptive Spectrophotom- etry in the Crystal 125

APPENDIX IV - Solution Acidimetry of Alkali Halide Crystals .... 127

APPENDIX V - Plastic Deformation of loni c FCC Crystal 132

APPENDIX VI - Crystal Growth and Evalu- ation of Metal Fluorides ... 134

REFERENCES 138

— ■ ' ■" -

LIST OF ILLUSTRATIONS

Fi gure Page

1 Anion Size versus Mulliken Electronega- tivity (•) and the Hydration Energy (■) 3

2 Growth of KC1 From the Melt 11

3 Czochralski Apparatus for the Growth of KC1 From a 10-cm Crucible Under Controlled Atmosphere 12

4 Czochralski Boules and Bridgman Cylinders and Flat Ingots of Single-Crystal KC1 14

5 A 10-cm Diameter Bridgman Cylindrical Ingot of Single-Crystal KC1 Grown by the RAP Method 14

6 Mechanical Evaluation of RAP Single Crystal of NaCl (B20) and KC1 (B19) and PolycrystallineKCl 20

7 Test Specimens of KC1 , Flexed to Failure, Viewed in Polarized Light 23

8 Scrubbing of Metal Chloride Crystals 24

9 Harshaw KC1 Scrubbed (20 Hours, 6700C) With CCI4 Vapor MOO mm Hg) 26

10 Behavior of Slip Lines (110) in KC1 With Thermal Treatment 27

11 Wire Saw Cut KC1 Using 320 Grit Silicon Carbide 30

12 KC1 Boule Sections Seen Under Crossed Pol aroi ds 32

13 Photo Sequence Showing the Effect of One Breath Cycle Leading to Condensation and Evaporation of a Film of Moisture 35

14 Photo Sequence Showing the Effect of One Breath Cycle Leading to Condensation and Evaporation of a Film of Moisture 35

15 Water Vapor Effects on RAP Annealed KC1 Surfaces 37

VI

■J""-•■^- .. ■ ■■■-■-■-— ^ ' ^■..^-.-^-«-^■t..1 .>.^....,...^ -..-■.,..-■.-■■. ■■-■,— ^w*..^.^-- ■■ . .-;.■■..■■■ ...■■■...-. ., .■..■, -.. -f*-^- ■'■■■■'■■■ .^.u.,^..*. ■....^.V„..^„..^,.:-.,- -..^. .._.■■■ ...t,^ , - , , , ,. . . . ...^-■■■-^„^.^■■..1._-.....l....J.v|.r r iritliM

.M« .iNAPVIinF!" ."■' <- 'fli."» •'•.' Trrrmi r^prw^A-i'-^MH- ■tP"!,ip U k lUJWn.HlMIPVF.IWMlftW I |..MJJ.JW[»!IM.,«l,,P! IIHI'.' M^V^MMi»wj|-iM

F- jure

16

17

18

19

20

21

22

23

24

25

26

27

2 8

29

30

31

32

33

Surface of Polytran Disc No. 77

Polytran Surface Before and After Scrub

Alkali Metal Chloride (KC1) Technology

Optical Absorption of p-Ni t rophenol Dye in Buffered Aqueous Solutions

Dependence of Percent Density of KC1 Disc on Pressure at 250C

Hot Pressed KC1 , 2.54 cm Diameter and 0.62 cm Thick, From Scrubbed and Unscrubbed MCB

Influence of PbCl2 on Growth From Solution

Solution Grown NaCl Crystals

Solution Grown KC1 Crystal

Infrared Transmission of Solution and Melt Grown flaCl

Double-Layer Anti ref lecti on Coating Design Diagram for ZnSe

Absorption per Surface in As2S3/ThF4 Anti ref lecti on Coatings on KC1 Versus Ma^ terial Properties

Theoretical Absorption per Surface in ZnTe/ZnS Antireflection Coatings on KC1 Versus Material Properties

Spectral Response of As2S3/ThF4 Anti- reflection Coated KC1

Spectral Response of BaF2/ZnS Antire- flection Coating on CdTe

Scanning Electron Micrographs of Polished KG! Surfaces

Scanning Electron Micrographs of Window Finished KC1 Surfaces

Laser Calorimeter Schematic,

Page

40

41

43

54

60

63

67

68

69

70

74

76

79

85

89

95

97

. 110

vn

-■—■- - - -- -- ■ - •■"--^ ■--■ —-■- --— - - ■ - ■ - - ■

11' ' ^^mmmm miiiLiii 11 "' ^mmmmia^mmmmmommimm

Fi gure

34

35

36

37

Page Optical Absorption of KC1 (Sample 1) at 10.G um Versus Thickness TIS

Optical Absorption of KC1 (Sample 2) at 10.6 Mm Versus Thickness -iiq

Transmission of a KC1 Crystal Specimen Polished With Linde A in Ethanol (Curve 1) and Then Polished With Linde A in a Glycerine H2O Solution Saturated in KC1 and pH < 1 (Curve 2) 123

Slip Planes in an Ionic fee Crystal -133

Table

I

II

III

IV

V

VI

Vi

VIII

IX

LIST OF TABLES

Analysis of KC1 Crystals, Modulus of Rupture S, and Absorption Coefficient at 10.6 um (ß)

Effect of Annealing and RAP on the Mechan- ical Properties of KC1 Crystals

The Value of ß in KC1 Secti.ons- Versus Cumulative Time in Heat Treatment :

Emission Spectrograph Analysis of KC1 Powders and Crystal Specimens. . . .

Impurity Content of Vitreous Carbon and Poco Graphite

Impurity Cleanup in Molten KG- der CC1

Theoretical Designs for Anti ref lecti on Coatings on KC1

Antireflection Coatings Without Low Index Fl uori des for KC1

Theoretical Designs for Anti refl ecti on Coatings on ZnSe

Page

15

28

33

44

45

46

77

77

81

vm

L.- ^r. _ 'fc, '.ita .. . -"i'J-"" Ik. ■ -— ■.■-—!.^J-^....-.-' ^.,,^.-... ...-^ ..,^.... ■ , ■ .-^^.■i.-i.w..,...,:.■.-■--.....i......■^.t-.-ji

Table Page

X Theoretical Designs for Anti reflection Coatings on CdTe S3

XI Properties of Passivating Coatings on KC1 Windows 33

XII Coating Results on KC1 37

XTII Experimental Anti ref lecti on Coating Results on CdTe Windows 07

XIV Pulsed Laser Damage Results gg

XV Absorption Measurements on KC1 Samples 93

XVI Absorption Coefficient at 10.6 ym Versus Humidity and Temperature Exposures -JIS

XVII Comparison of Models I and II in KC1 120

XVIII The Value of a in A^ (m - 0.67 g; L = 0.?7 cm) Versus Surface Treatment 122

XIX Comparison of Metal Fl uori de Crys tal s 126

ix/x

■■

SECTION I

INTRODUCTION AND SUMMARY

The Ilalide Window Materials Technology Program is a

continuation of work started on the Advanced Mode Control and High-

Power Optics Technology (AMCIIiPOT) Program. The AMCHiPOT

Program is intended to advance the state of the art in high power optical

components and high-power laser mode control. The AMCHiPOT Pro-

gram includes: advanced mirror and grating development, windows and

coatings, and mode control. The results of hoth the advanced mirror

and grating development work were reported in final technical report

Volume I, (AFWL-TR-72-152). Volume III will document mode control

findings, while the present volume (Volume II) includes results on the

windows and coatings studies.

The object in the study of window materials was to upgrade the

technologies of purification, crystal growth, and surface preparation of

halides for their use as windows for high-power lasers. The approach

taken was based on our hypothesis that the values of material param-

eters employed in the earlier evaluation of figures of merit were not

intrinsic to the materials but reflected the status of halide materials

technology. Our results now allow us to assert the validity of this

proposition.

At the start of the program, the optical absorption at 10.6 [j.m,

P, for KC1 crystal (Harshaw, window grade) was 0.007 cm" . Currently .1

P s 0. 0005 cm can be obtained in a reproducible fashion. Initially,

the yield point was Y = 300 psi and the rupture strength, S = 640 psi.

(Ref. 1); Hughes Research Laboratories now routinely achieves

Y ~ 1000 psi and S = 4000 psi. Preliminary evidence shows that

further improvements in optical absorption and mechanical integrity

can be expected from technological advances in surface preparation.

The improvements are noteworthy in that they were realized in the puri-

fied material, and were superior to the improvements obtained earlier 2

by bivalent metal ion doping. Related to the difference in the two

approaches to upgrade (Y, S) is the observation that doping exacts a 3

tradeoff in ß. Bivalent cation impurities stabilize the hydroxide ion 4 (OH ) and cation vacancies in alkali metal halides. Also, because of

the extremely limited Eciubility of such dopants in the pure solid, the

achievement of bulk homogeneity over a large piece presents a severe

constraint; an increase in optical scattering through precipitation is a

constant threat.

The emphasis in materials purification by reactive atmos-

phere proci. ssing (RAP) (see Appendix I) is anion purity. The

nonmetal impurities, being much larger than the metal ions, affect

the lattice array at concentration levels of a part in 10 or even lower.

Regarding the ion as an incompressible sphere, the halide anion

accounts for a major fraction of the molar volume. Consequently,

the anion determines generally the closeness and symmetry of

packing; its impurities exert a greater influence on lattice vibrations

and, therefore, affect the trystal's IR transmission, structural

integrity, and transport parameters.

Among the anion impurities, the hydroxide ion (Oil ), a pseudo-

halide, is probably the most obstinate because the reactivity of the radi-

cal is very close to that of fluorine as shown in Fig. 1. From the

standpoint of optical transmission, the O-Il vibration is active in the

3-fjLm region and the O-M at the in-fj.m region. The substitutional OH"

dipole couples with the dipole consisting of a bivalent metal ion impurity 4 5 6 and a metal ion vacancy. ' ' Through such couplings and the tendency

to diffuse to the surface, the Oil" can effect a degradation of the mechan-

ical behavior of the alkali halide. Our surface-stability studies indicate

tl at the presence of Oil on the alkali halide crystal surface renders the

surface more hydrophilic and, consequently, less resistant to surface

corrosion by water vapor.

Because of its reactivity and the omnipresence of its source,

H^O, in all steps of processing, the substitutional entry of OH" by hydro-

lytic exchange is always a threat. To suppress the substitutional entry

of OH in the growth of chlorides, the critical process parameter in

RAP is the ratio of partial pressures, P.. n/P „ (see eq. (2),

^

126^-4

Ott

ev

Fig. 1. Anion Size versus the Mulliken Electronegativity (•) and the Hydraticni Energy (■).

..- ■... ■^;.- . --..^■■-,..:.-. ,_..!..-_.. ..■.-■-.-. Ji.-.^^.l||ti ..;)■■.. ..

Appendix I), which should be taken as close to zero as is compatible

with the operation. Consideration of this process parameter shows why

the current popular approaches, processes based on atmosphere control,

are far from achieving the effect of RAP. The use of vacuum or an

inert gas is not wholly effective in excluding OH from the solid because

PJJ,Q is never zero while PT,ri is essentially zero. The same situation

holds in ihe use of chemical getters in the condensed phase (melt), e. g. ,

PbF, in fluoride melts to remove the oxide as the more volatile PbO.

The latter process involves a tradeoff of a cation impurity, Pb , for

an anion impurity, O , which may have an adverse effect on ß (Ref. 3).

A reactive atmosphere process incorporates a reactive vapor

phase which reduces the process parameter, PH-Q/P^,-^, to a very

low value by suppressing the concentration of II O and enhancing that of

IlCl. One approach is to live with the steady-state outgas pressure of

PTT-^Q and apply a positive P ,... Unfortunately, the weighting factor to

the process parameter is the equilibrium constant of hydrolysis (see

Appendix 1) which can be larger than unit/ at the melting point. Also,

aside from the problem of corrosion, bottled 1IC1 gas has ordinarily an -4 -7

effective PJ_J_Q/PT _. 2: 10 ; the desired value is <10 . Such a dew-

point in the source is difficult to obtain and maintain in a flowing system.

A more effective approach to reduce the pressure ratio makes use of a

reactive vapor which produces IIC1 at the expense of II O. This optimi-

zation of RAP, leading to the choice of CC1 , is presented in greater

detail in Appendix I. The above problems lead one to appreciate that

further optimization of all processing steps, from starting powder to the

single crystal (congruent growth), into a one-step RAP was a very sig-

nificant development in this program.

Optical, mechanical and surface-stability evaluations all show

the advantages derived from the vise of RAP. With a dynanr^ flow one-

step RAP it became possible to start with a 99. 9% pure KC1 powder,

priced at <1 $/lb, and obtain the san-.e performance as with a 99.999%

pure starting material, priced at ^lO $/lb. Optimizing a one-step

RAP with the 99. 9% powder improved ß from 0. 0016 cm" to <0. 0005 cm'

- ^L-rr^r^n-^w^Mtaat^i^^^.. rr.^^- . _.... -_ ■ ..

One need not elaborate on the obvious significance of this result for the

production of large single-crystal ingots.

An earlier study by other workers reported a large surface

absorption at 10. 6 (im (Ref. 1). For KCl, the reported surface absorp-

tion was >80% of the total absorption for a I-cm thickness (Ref. 1).

The cause was thought to be the hygroscopic behavior. However, our

RAP treatment of the crystal surface, i. e. , topochemical exchange, did

not improve p. In addition, we showed that neither surface hydrolysis

(exchange of Oil" for Cl") nor hygroscopic behavior (surface uptake of

H20) contributed to the surface absorption of a KCl crystal with

P 0. 0042 cm , with measurements well within a precision of

i0. 0004 cm . Instead, our study indicated that fabrication damage

(surface preparation) can contribute to bulk absorption in a manner

which could be mistaken for surface absorption. These results are collected in Appendix II

Mechanical evaluations (Knoop hardness and rupture strength, S)

showed the benefit of a RAP treatment following fabrication of the test

specimen. Evidence was obtained showing this also to be the case in a

study of surface stability of KCl cleaved at (100), at humidities exceed-

ing 100 percent. At the end of the program, we felt the need for devel-

oping a milder RAP treatment for the more refined requirements of the

last stage of surface preparation; i, e. , prior to coating.

In the last quarter considerable effort was devoted to the quanti-

tative determination of the Oil" content of KCl crystals, commercial

versus the RAP-grown. The primary method, acidimetry, is a more

reliable alternative to the optical absorption methods. Unfortunately,

the testing is destructive and does not differentiate between the surface

layer OH" and that in the bulk. At the end of the contract, the problem

had not been resolved of defining a reference material where C, the mole

fraction of OH , was zero. The measurements support the expecta-

tion that C < 10"5 for RAP-grown KCl and, if such was the case, then

C = 2. 2 x 10" for the Harshaw KCl. Further background material on the subject is given in Appendices III and IV.

- — — ^J—____„_^-_„^-^_—,_ ^_^— .

The study of noncongruent (solution) growth was not intended for

large crystal production. Theoretically, the method is capable of pro-

ducing crystals with very low Oil" content. Conceivably, crystal speci-

mens with known OH' content could be grown which would serve as ana-

lytical standards for melt grown materials. Although clear crystals

were grown, the study showed that the probability, per unit volume, of

producing a fairly occlusion-free crystal was not low enough with growth

recipes which included those given in the literature. Our best solution-

grown XaCl and KC1 were inferior in ß and in transmission from 2. 5 to

25 \im to material melt-grown by a one-step RAP.

The objective in hot pressing was to press mixed metal halides

from the powder to effect good bulk homogeneity and improve mechanical

strength. The study began with a one-component system (KC1) to

acquaint ourselves with, and solve, problems of the technology: powder

preparation, compaction behavior, pressure versus temperature trade-

off, interface corrosion, etc. Our initial comparison of the pressed

disc with the single crystal showed a ß which was one order of magni-

tude higher. The problem was aggravated by marked interface corro-

sion resulting from the use of powder that had a preliminary RAP

treatment. Towards the end of the program, chemical compatibility at

the interface was achieved by pumping out the reactive gases from RAP

before pressing. C nsiderable improvements followed in the optical

transmission and in the mechanical strength of the pressed disc. The

applicability of hot pressing to IR window materials for high-power

lasers hinges on solving the problem of ß. From the experience gained

in the other technologies, we expect dramatic improvements /rom the

development of a dynamic one-step RAP in hot pressing.

We carried out a low-level effort study on the crystal growth and

evaluation of metal fluorides as window materials for the 3- to 5-fj.m

region^ Both ß and the Knoop hardness proved to be useful measures of

the efficacy of RAP with He + IIP at a dewpoint of PII2O/P ~ 10''">-

The use of ß as an index to anion puritywith respect to OH" was based

on the activity of the metal-to-oxygen vibration at the lO-[im region,

■ -■ "-- ' ■ -— ' ■• • ■- -- - ^ ~- - . ■■ --^- ■■■ i , E - ■ ..-...■- --' -

w

which, in turn, was a measure of the absorption at the 3-|jm region

arising from the oxygen-to-hydrogen vibration. The observed changes

in p could have been consistent with larger changes in the optical

absorption at the 3-|j.m region where the absorption cross section is

lirger. Unfortunately, this correlation was not demonstrated as there

was no apparatus available to measure low-level optical absorption in

the 3- to 5-|j.m region. The Knoop hardness was a qualitative indicator

of the effect of RAP on the mechanical properties, but more direct evi-

dence was obtained after the contract ended: The maximum strength

reported for CaF-, is S000 psi at Raytheon (Quarterly Report December

1970, ARPA Order 11 SO); our RAP-grown CaF-, gave a modulus of rup-

ture 12, 000 psi. The study, reported in Appendix VI, shows that the

advantages of RAP are best realized if the procedure is applied begin-

ning with the conversion step to the fluoride powder and maintaining the

specific surface to reverse hydrolysis prior to crystal growth.

The results of optical, mechanical, and surface-stability evalua-

tions of metal halides at the Hughes Research Laboratories support the

main thesis. The intrinsic limitations of alkali halides as windows for

high-power lasers remain to be defined by the future development of

materials technology.

The objective in the window coating task was to develop anti-

reflecting/passivating optical coatings for candidate high-power IR

laser windows. Major emphasis was placed on the goal of achieving

0. 1"'- or less absorption per coated surface for the optical coatings.

Theoretical designs and experimental results for coatings for alkali

halides, cadmium telluride, and zinc selenide are reported, with the

major emphasis on coatings for KC1.

The performance of the optical coatings on KC1 is greatly depend-

ent on the surface finishing operations performed on KC1 prior to coat-

ing. Results of experiments on surface finishing of KC1 are presented

and the optical absorption in the KC1 and the coatings is discussed rela-

tive to the surface finishing and handling operations.

At visible wavelengths, the availability of low optical absorption

dielectric materials has made it possible to produce required window

- - ^______J^„__— - -

coatings. One of the major problems that had to be overcome was

optical beam losses from scattering caused by the coating texture. At

10.6 (im, low absorption dielectric materials have only recently begun

to be available; however, optical beam losses due to scattering may

not be a major problem and at this longer wavelength, some texture in

the optical coatings may be tolerated without major contribution to

scattering losses.

Window materials that show promise for use in the near infrared have refractive indices that fall in the 1.5 to 3.5 range. Anti-

reflection coatings for the low index materials serve the dual purpose of

improving the optical performance of the windows while simultaneously

protecting the halide window surface from environmental degradation

(since one of the promising candidates among the optical window mate-

rials for use in fhe infrared, KC1, has appreciable solubility in water).

For window materials that have good optical characteristics as

a result of their low optical absorption losses but which have indices of refraction greater than 1.5, the lowered optical transmission due

to the high reflection from the window surfaces makes the use of anti-

reflection coatings mandatory.

The surface finishing and coating technology, established to date,

has produced the following results for 10,6-fjLm absorption in our anti-

reflection coatings:

KC1

Two coating designs for KC1 were investigated. One design is a

two layer coating of As-,S, and ThF. in which we have achieved 0. 19%

absorption per surface for the coating; the second design is a two layer

coating of ZnTe and ZnS which has produced coatings with 0. 05% absorp-

tion per surface.

CdTe

Antireflection coatings for CdTe were produced using a coating

vhich consists of two film layers of I

have achieved 0. 06% absorption per surface.

design which consists of two film layers of BaF^ and ZnS in which we

„........_ ..—_..■—.■^--.... ..■.. ■ . ■...~^.-L~^ i~. ..—.....,- .L...-,.. ,..., '--■•-'—-'—-^-iiitiiittitiifi'JIiiir^HWu-tm i i TM i i ■ - Mm-'1! -'■■- - . ....—■■...-—^.■^J-.- - --■^- ■■■.m,. i^n,, tiiM<trninl linllhMl

The ambitious goal for our coating program, which was projected

at the first laser window conference at AFCRL in October of 1971, was

to achieve less than 0. 1% absorption per surface for window coatings.

The results, achieved to date, are impressive when one considers that

in October of 1971 our state of the art in antireflection coatings was

absorptions per surface in the 0. 5% to 1% range.

Our ability to achieve the coating program goals was based on

the following combination of capabilities available to us at Hughes

Research Laboratories:

• Availability of high quality single crystal KC1 and CdTe window materials grown at Hughes.

• Surface finishing technology for KC1 and CdTe which provides minimum degradation to the intrinsic optical absorption of the bulk window materials.

• Availability of processes for preparation and purification of AS2S3, /nTe, and ThF4 starting materials used for film preparation.

• 10.6-|j.m laser optical monitor for precise thickness control of optical film layers.

• Complete and comprehensive computer design and analysis programs, which include optical absorption in the films, so that theoretical designs and experimental results can be compared.

• 10. 6-|j.m laser absorption calorimeter for measurement of absorption in coated windows, window substrates, and for determination of absorption indices for optical films.

The important requirement for future coating work is to continue

efforts at improving window materials, window surface finishing and

coating technology so that windows when subjected to 10. 6-iim high

power incident laser fluxes will not fail at damage thresholds far below

the intrinsic damage thresholds for the bulk window material.

~ *■ ^ '- '- ■-' -...- .,.....—^ -I-.W, —-. . ,. .. ■■ ..,-■■.

SECTION II

RESULTS AND EVALUATION

A. WINDOW MATERIALS

1 • Crystal Growth and Evaluation of KC1

The technology of single-crystal congruent growth from the

melt under RAP was studied for NaCl and KC1. Greater emphasis

was given to KC1. Since the technologies for the two materials are

quite alike, they will not be discussed separately.

The study began with the design and construction of a

Czochralski apparatus for 1- to 2-cm diameter crystals (Fig. 2) which

could be adapted to a RAP operation. The apparatus provided speci-

mens for optical evaluation. It was also used to study interface com-

patibilities (corrosion) and transport problems (impurity pickup)

accompanying RAP. Optical evaluation showed that radial inhomoge-

neity of the refractive index; i.e., lensing, was not insignificant.

Consequently, a larger Czochralski apparatus with a 5-cm diameter

capability was constructed (see Fig. J).

The first few runs with the large Czochralski 'apparatus showed

Poco graphite crucibles to be permeable to molten KC1, behavior which

led to crucible failure during cooldown.^ Vitreous carbon crucibles

from !' | with Corp. were used to grow several T-cm diameter

boules. ! h, se boules showed elongated void.-. Bubbles formed at

the growth interface and nucleated these regions of negative growth.

It was observed earlier that the gas atmosphere in RAP manifested

a high solubility in molten KC1 and lowered drastically the distribution

coefficients of impurities. This feature was, of course, an asset to

scrubbing in purification but, in turn, caused a larger flux of gas

(bubbles) during the change of state from liquid to solid; i.e., crystal growth.

10

^ -- . -....^ .,-.. ,. „.—^ ..--■.■..^^..^.J. ...... , - - ■■ ■ , , , -■-■-■■■■ - -

;iiriiii06E1<

••.iulK COWl«

:. , ;. «HMIÜVJ

3 ri1

0 7?J

(i i";'

"^ 0.136 0,120 (HID 0 152 ii'.Mrs U.3S (I 337 0.45 0.521 ii S353 o.ni

n 7Wil MtWtUVJ 0.1120

0.1 i.noil I 220 i 3505 ; 30

Fig. 3, Czochralski Apparatus for the Growth of KC1 from a 10-cm Crucible Under Controlled Atmosohere.

12

. J. ..^..^U-l.-^-/..- ' ■ - ■ ' '

An internally funded study on the technology ol" scale-up in

RAP crystal growth paralleled this program. RAP crystal growth was

developed in the vertical Bridgman mode, a geometry more favorable

for coping with the problem of gas at the growth interface. The tech-

nology also allowed for the optimization of interface compatibilities,

a problem accompanying the use of reactive atmospheres. Another

advantage of the method was its adaptability to the growth of flat

single crystals. Figure 4 shows a collection of single crystal KC1

Czochralsk^ boules ranging from 1 to 5 cm in diameter and Bridgman

cylindrical ingots and flats up to 5 cm across the cross section. The

elongated voids described previously are "seen extending through the

bulk of the large Czochralski boules. Figure 5 shows a 10-cm

diameter Bridgman cylindrical ingot of single crystal KC1 grown by

the RAP method.

The optical absorption at 10.6 \im, (3, was measured with an

adiabatic calorimeter. The measurement technique and the apparatus

employed are discussed in Appendix II. Surface hydrolysis and

hygroscopic behavior were shown to have a negligible contribution

to surface absorption; however, the contribution of fabrication damage

was found to be significant. If such a contribution was not separated

from the bulk absorption, the extrapolation of power absorbed to zero

sample thickness gave an intercept which could be misinterpreted

as a contribution from the surface. Also, a consequence of the appar-

ent surface absorption was an exaggeratedly low bulk absorption

derived from the slope of the plot of power absorbed against sample

thickness1'.5. However, legitimate contributions to surface absorption

exist. An instance such as this was shown to arise from the ingredi-

ents employed in surface polishing. These earlier developments are

collected in Appendix II.

Using KC1 crystals of different purity grades, no correlation of

ß with the metal ion impurities was established. Table I lists the

results of emission spectrograph analysis, modulus of rupture (S),

:::The ambiguity in the value of the slope arises from the scatter in the experimental values of (x, y). The least-square fit to y = mx+ b yields a smaller value of m than the fit to y = mx (cf. Fig. 35, Appendix II). When there is no surface absorption, the observed absorption must all be accounted for in the bulk.

13

mmmmmammmtmmmmi^^mamm^mmk^^tm

W5

Fig. 4. Czochralski Boules and Bridgman Cylinders and Flat Ingots of Single-Crystal KC1.

Fig. 5. A 10-cm Diameter Bridgman Cylindrical Ingot of Single-Crystal KC1 Grown by the RAP Method.

14

■■Mt ÜifciVi -i^^»^-.■■•■— .^.--..L.^..

T 1 ^f T* O I —

arshaw

Cone

0. o

ooc

0. 0011

0. 0005

0. 0009

0. 0012

0. 0007

ro O O

d

00 in

O O

ffl i V V O 00 Q p -< -H

1 '■• Z Z

"<t vD v£) O -4 O O o o O O -H o

o O O o ro o

P-H

o o J2 (U O O o O O o O ^^

m <u • • , Tf ^ o O o o o d o o o o ffi V V V V V O O

w 0 Q Q Q p M z z Z Z z 1 ^ c 3 £ IT

0) ro QJ i o ro vO r~i ÜT

s. Modulus of

ficient

at 10

. 6

L

Zo

fined

. 0001

0001

0003

0008

001

0007

O O 0

005

0000

i i

I-H

«(5 o O

V

Q

o O O O

V v

Q Q

d d d -H

w z z -1

< H

KCl Crystal

orption Coef

JM

?. 999%)

0. 000049

0. 0001

0. 00039

0. 00032

0. 001

0. 0007

ro O O

d 0. 00082

0. 00005(3)

i i

# & (>

O A o i V V V V ■H o

*< 1 0 Q cd p m ',« "0 z Z H z u

- i I a- (NJ O

< - r^ r-, o o

ro ro |^ O O -H o

o ro

in o i—t o

c

u ^ o o O O

O O O O o O O O § o o

o o ? Vl

2^ d d O O O O o o o O O (U

.

cr V Q Z

V V

Q P z z

■w ro Tf O^ (M

-H

c 0 N

•« cn CO

FH

1

£ 0)

3 W) « <« a> .^ u a | x;

U < S U h 2 •H V3 Ö M-

H

1

in sO in H

15

.. ..L .-■i--:^^.-^^^,. ■ ■ . .

and ß. The first three samples from the left, identified by the source

material in the column heading, give the analysis of the tips of

Czochralski crystals grown at Hughes Research Laboratories (IIRL).

The other two are Harshaw specimens sampled from two different

regions of a Stockbarger crystal. The metal ion impurities are

grouped according to the most likely valence assumed: +1, t2, and +4.

No correlation of S and (3 with groups I and IV impurities can

be drawn. However, with bivalent impurities, the Harshaw cone

niaterial is a factor of two to three larger in total content (-oO ppm),

which may account for the higher S compared to the heel section

(Ref. 2). Yet, according to the results with MCB specimens, an

increase in the level of bivalent impurities is not necessary to attain

S « 10 psi.

Bivalent impurities can lead to a higher \i through the stabili/.-

ing effect of the impurity on OH" (Ref. 7). The effect could account

for the increase in ß of KC1 with Pb" doping, (Ref. 3) although PbCl,

itself has a low ß(0.0020 cm" ). On the assumption that OH" was

not excluded in the PI) doping of KC1 I Ref. 3), the model leads to -19 2

an absorption cross-section at 10.6 |xm of, a < 10 ' cm . The

model would also explain the tradeoff between ß and S in the heel and

cone sections of the Stockbarger ingot of the Harshaw specimens.

The MCB result shows that there need be no tradeoff in ß for

a high S. Although the three HRL specimens show the lower ß, no

correlation with impurity levels can be drawn. The molar sum of

the bivalent impurities is constant for the three, corresponding to a _5

mole fraction, x = (2.66 ± 0.03) x 10 . These results are consistent r 7 1? 1 Q 7

with the interpretation that: C < lO-3 and 10" cm < ir < 10"' cm ,

If ß is not metal-ion impurity limited and RAP growth is capable of -7 -20 2 achieving C = 10 , then from cr= 10 cm the value of ß limited

- I .22 by OH (s) is 10 x 10 x 10

the anion density in KC1.

-20 -5-1 10 cm , where 10 cm is

Supplier code of KC1 source powder: MCB = Matheson, Coleman, and Bell and JM = Johnson Mathey.

16

-■■-—■—^■--' '-■■ ■■■ - ■ ■ ■■■■ — - - - - - - - - -'--' ....■..^-J.W.

The value of ß, OH~(s) limited only, is two orders of

magnitude smaller than the average for the three HRL specimens, -4 -1 which show a standard deviation of Aß =5x10 cm . It is neces-

sary to locate the source of Aß as ß is at most 3 Aß. Annealing

experiments show that a source of Aß is the bulk damage brought

about in surface polishing (see Appendix II). These results indicate

that ß is now limited by the technique of surface finishing instead of

materials processing.

The above conclusion is supported by the results of evaluation

of 5-cm diameter Bridgman ingots. Thin disks, ~ S-mm thick, were

taken from each end of the cylindrical section: ß = 0.00082 cm"

at the region which crystallized first and ß - 0.0014 cm" at the

later stage of growth. The intervening cylinder, L = 11.2 cm,

gave 9 ^ 0"1) transmission uniformly over the face within 1% variation.

With no absorption and scattering losses, the theoretical transmission

is 93. 55%. Thus, the value of ß over the cylinder is 0.0035/11.2 =

Ü. 0003 1 cm" . The end faces of the cylinder were wedged to avoid

interference effects.

Similar results occur in the case of NaCl. Our best value

on melt grown Harshaw NaCl is ß = 0.00 5 cm" . A nonoptimum RAP

on a C/ochralski boule (sample NA-9) yielded ß = 0.0018 cm' for

L = 2.69 cm. When this sample was sectioned further with a wire

saw, the limiting effect of surface polishing is seen as follows:

ß = 0.0016 cm"1 for L =0.85 cm, ß = 0.0023 cm"1 for L = 0.80 cm,

ß = 0.0025 cm" for L = 0. 52 cm, and ß = 0.0037 cm" at L .= 0.47 cm.

However, with an optimum RAP Bridgman ingot, a cylinder with

L = 12.0 cm showed a uniform transmission of 92.0"() which was the

theoretical value for no absorption and scattering losses. Yet, at

the region which crystallized first, ß = 0.0028 ± 0.0005 (3) cm"1,

and that which was last, ß = 0.0030 ±0.0005(5) cm" . Obviously,

these two measurements do not reflect the intrinsic value of the

material but the limiting effect of fabrication. Otherwise, ß - 0.003 cm'

over L = 12.0 cm would have shown a 5.6% discrepancy below the

theoretical transmission. Again, the end faces of the cylinder were

wedyed.

17

- -- — -

For the measurement of yield (Y) and the modulus of rupture (S),

Instron equipment was employed to carry out the flexure test to failure.

The crystal, in the form of a parallelepiped, was mounted as a simple

beam resting on end supports with force applied as two equal concen-

trated loads at the one-third points (lour-point loading!. A slow

loading rate (Ü.005 or 0.002 in. min" ) was used to give reproducible

readings of S. At the beginning of the study, the parameters of commer-

cial KC1 crystal were: Y = SOO psi and S = 640 psi (Ref. 1). As seen

in Table I, S =10 psi can be achieved with no tradeoff in ß in the

undoped material.

Parallelepiped specimens were prepared by cleaving at

(100); either llarshaw or Optovac KC1 gave Y = ^00 psi and S = 700 psi,

values in agreement with the parameters given above. With a nonop-

timum RAP procedure and the same method of preparation of the test

specimens, the IIRL specimens yielded, Y -: 400 psi and S = 9^0 psi.

The preparation of the surface of the test specimen (llarshaw

and Optovac KC1) was also shown to play a significant role in the

results of mechanical testing. Using cleaved specimens at (100t, the

measured average (S ) and the standard deviation (AS) showed that " av uniformity in test performance improved with polishing (Linde A in

alcohol) but the average strength decreased; i.e., AS and S were a a » av

lower. The lower value of AS indicated that polishing removed the

shallow microcracks; the lower value of S suggested that polishing a v may have paused some of the cracks to propagate into the interior

faster than the surface material was removed, and/or surface'recrys- Q

tallization occurred after polishing. Thus, without polishing,

Optovac specimens prepared by cleaving yielded S' = 4 30 ± 50 (2) psi

and, if followed by water etching, S = 8 50 ± 360 (11) psi.

Water etching removes an- 0. 5-mm layer of the material. The sample is swished for 1 to 2 sec in water, followed by a rapid alcohol rinse and hot air drying to avoid surface recrystallization.

18

, .. _ . . inii:ni-»i^ii^tdi« ittmäimik .■■,.■;. .^^--^..i..-.!—!-.. . .. . - •—■■■■■- ^-•- ■>■ — ■:;'—'^'—■■■'.> ^ ^_ ^U-^.- ^t..^.'.^

~™!-J''?WTO!WWWWBPIlyM11 i I S»iffBBP«!!iHPWl^P»WW1WRPP*W1Blin»^SWl»JW^^^I?WiffP^^*IWifWW!ipiPW(^^

The nbove results suggested an examination of annealing

procedures as a means of repairing the damage on the surface and in

the interior caused by cleaving. Harshaw KC1 specimens were used

and the following results were obtained:

When cleaved only,

When cleaved and lie-annealed,

When cleaved and RAP-annealed,

S = 640 psi av

S = 450 psi av r

av 970 psi

Annealing consisted of a 20-hour exposure at 670 C to a pressure of

100 to Z00-mm Hg of either He or CCL^.

The effects on the yield point (Y) of the generation and treat-

ment of (100) surfaces of test specimens were studied. Since damage

from the surface into the interior followed cleaving (Appendix II)»

cutting with an 8-mil diamond-impregnated copper-clad steel wire

was examined as an alternative. Cleaved and polished Optovac KC1

gave Y = 450 ± 50 ( M psi, while the same material cut and polished

yielded Y = 580 ± 40 (4) psi. When polishing was replaced by water

etching, the difference was even less, Y - 400 ± 50 (111 psi versus

Y = 410 ± 60 M) psi, cleaved versus cut, respectively. The lower

value in the Y suggests that water etching removes a work-hardened

surface. This proposition also implies that a cut surface work-

hardens in polishing better than a cleaved surface. The opposing

effects of cleaving followed by water etching should be noted; A

dramatic increase in S and a decrease of Y from the work- av av hardened value to what may be intrinsic; i.e., a widening of the

plastic region.

Test specimens were fabricated from single-crystal RAP

Bridgman KC1 and NaCl and polycrystalline KC1 (Polytran) by cutting

with the wire saw and polishing with Linde A in alcohol. The Polytran

specimens showed a yield of Y = 580 ±20 (3) psi and the HRL speci-

mens gave Y = 660 ± 120 (7) psi. A summary of the results of the

(Y,S) measurements is shown in Fig. 6. It is seen that in the case of

19

■-- •"■— - -

r

I68I-MRI POLYTRAN DISK BRIDGMAN INGOTS

YIELD(Y)/RUPTURE(S) IN PS!

IRSI- I2RI

560/7801 580/7001*

KC/.BIS

610/2200 NOT OBSERVED/MOO

610/740

780/2510 690/3740 340/2930 850/3420

Arnow GIVES ORIENTATION OF SPECIMEN WITH RESPECT TO FORCE APPLIED (INSTRON)

NoCi,B20

NOT OBSERVED/1200 I 680/1100

2l80/2570( CENTER) 980/1930 (SIDE)

1200/2055 NOT OBSERVED/1420

<I00>

Fig. 6. Mechanical Evaluation of RAP Single Crystal of NaCl (B20) and KC1 (B19) and Polycrystalline KC1

20

- - .,...MJ^J^^ia.^^niM»j-_»Mi^a^^^.1-^J.i^^ „ -"• -'^ '

■" ■ i ii j ii i i *mm,m,. m m i i\tmmiimmmm*mmmmmmmr^^*^**m^~<^*'^*^^^*~im m w*^^imma^*~r~mm^mi^^m*m*^*r

polycrystalline KC1, Y increased slightly while S remained the ^ 7 ' av => / av

same as the earlier value; therefore, the plastic region was narrower.

In the case of single-crystal KC1, Y increased by more than 100%

and S increased by as much as 500% above the earlier values. The av ' net effect was a wider plastic region, without the use of water etching.

In fact, the S values in this case exceed the S obtained in the av av

earlier specimens (Czochralski, nonoptimum RAP) by the use of

water etching.

Similar increases also occurred in the case of single-crystal

NaCl (early value of S = 570 psi) but, as seen in Fig. 6, the plastic

region is narrower than in the case of single-crystal KC1. At times,

no yield is observed up to failure.

The growth direction of KC1 ingot B 19 was found to be close

to < lll^ The observation suggested a search for an optimum direc-

tion in (Y,S). For window application to high-power lasers, the

optimum in Y was preferred over that of S if the optimization could

not be effected simultaneously. Consequently, water etching of the

wire-saw cut specimens was employed. The results which follow give

the test specimen axis along the long direction, the test direction,

and Y:

< 100>, < 100>, and Y = 390 ± 30 (7) psi

<110 >, <111 >, and Y = 500 ± 50 (4) psi.

< 111>, < 111>, and Y = 530 ± 50 (4) psi

From these results in Y and those of S (Fig. 6), it is quite likely

that the growth direction along < lll^ provides the optimum in Y and S

(Appendix V). This growth direction is realized in spontaneous

nucleation.

Although the program had ended, it was felt necessary to check out this feature with a commercial crystal. Optovac KC1 specimens were prepared by the same procedure, i.e., wire-saw cut followed by water etching. With the long bar axis at < 100> and test direction at < I00>, Y = 410 ± 60 (3) psi; with the long bar axis at < 111 > and test direction at < 111 >, Y = 600 ± 1 50 (7) psi.

21

■--■'■•-—■■■— --■ ■ ■- .----^^ .-> .--.-■■.-.»■-■t...-i-....1IIM1lf.l,, r. .■..,.,.. ......■■...,■;,.■. ■■.■^.....-■■^ . "" ._._.•* -WSä

-ir—rr--" "»-"■ ^—.-~ fm., n'n.or . jii, ^(MiLi-JJUJiH, I (mi'.jJi.(!M"l".:«rwl^,IAB!PWl'HH*ff'"Vl»W"

Figure 7 shows the different failure patterns revealed by

polarized light. Figure 7(a) shows KC1 specimens with the (100)

surfaces generated by cleaving and tested at < 100>. Starting früm

the left, the first seven are llarshaw crystals, the next five are Optovac,

and the last lour are IIRL. With either llarshaw or Optovac, Y = 300 psi

and S 700 psi; the IIRL specimens, Y = 400 psi and S : (.H0 psi. The

test direction was applied from rij^ht to left; the density of slip planes

at (110) reveal the load points, and the failure occurs at (100).

Figure 7(b) shows the wire-saw cut specimens (polished)

from R lf). The test direction, from ri^ht to left, is < lll^ The

long bar axis of the two at the left and the two at the right are perpen-

dicular to ' I 1 1-■ while the three at the center are parallel. The

specimens store a larger amount of strain energy in the testing and

break with a louder report than the specimens of Fig. 7(a). The

failure regiqn, where the strain pattern is washed out, when observed

in the test direction shows a wedge of two (100) planes.

Early in the program, compressive strength tests on Optovac

KC1 revealed yield strengths of ZZ5 to 2 30 psi and a rupture strength

of 2410 psi. Plastic flow occurred only in one of the < 100> direc-

tions, while the other < 100> directions remained constant.

2. Topochemical Exchange (Scrubbing)

In this study we established a RAP procedure for the surface-

layer treatment of KC1 crystals at temperatures below the melting

point. The arguments given in Appendix I for the choice of CC1 ,

as the exchange agent to reverse hydrolysis apply to the present case,

The apparatus employed He as the carrier gas or was operated as a

closed system under a vapor pressure of CCl^, as shown in Fig. 8.

It was observed that surface hardness (Knoop) increased after

scrubbing with He I CGI .(g) but not with He alone. Hence, Knoop

measurements were used to study the depth of penetration of

scrubbing at various temperatures and dwell times.

22

M^, u. „^ ■■- -■-— -- ■ ■- ■■ -' - -■ -

■■■I m •>•'• ■■■!—>■ numtmmm*** » \i * im «i « «.n n tu i unn n m t^t^^mm^^mmmmmm^nfmt« i^^^mmmt^^m^mnmimmmiiimn

^' w '

' ' V / \ \< /

<■ \ \ \s >

\ 0 /

\ \ > > « Xf^ V . vv . X V B ^v , N A \ \ / } -\r f

i / / / /

' 1 '> ^ * i-'i.

' % r v v / f ^^ l V ' ^ ^:

1 cm

(a) ( 100) surfaces tested at <100>

4 "4 3C 3C 30 2 2

(b) (111) surfaces tested at <m>

Fig. 7. Test Specimen':; of KC1 , Flexed to Failure, Viewed in Polarized Light

23

»-.•■■' -^ , -,.. -.. .- ^...-.^ ■:.,. :

"■■«W■"I■■■""■ ' I I Ml 1 I ■^^^|MPn>MW«iH4W«^P«WW«PV>^W*»W^^OTV^MWW" I 1W 1 «■^^^■^^»W!^W^WÄW»W^»»»«W^WWWWW>^B|^

Specimens were obtained by cleaving at (100). The hardness at

depth zero was taken on the as-scrubbed material. Then the scrubbed

material was cleaved from the opposite side. The depth of penetration

was measured away from the region where contact was made with the

cleaving tool. At 450 0C, no penetration could be established up to a

35-hour dwell time for KC1. A SOO-^m depth was achieved with 500 0C

and 15-hour dwell time for KC1. A comparable penetration was achieved

for KBr at 350 0C and 90-hour dwell time using CBr,. 4 A greater depth of penetration was achieved with an 8-mm thick

KC1 crystal, by a 20-hour exposure to 100 to 200-mm Hg of CCl.(g), in

a closed system at 670 0C. This is shown in Fig. 9. These conditions

of exposure were also found to be adequate for the removal of slip lines

in KC1 crystals, as shown in Fig. 10. These slip lines are at (110) and

are generated by cleaving and can be seen to a few millimeters depth by polarized light.

Table II shows the effect of annealing and the atmosphere

employed in the process on the mechanical properties of KC1 crystals

(Harshaw). The best improvements in mechanical properties are seen

to be associated with the RAP surface treatment of a 20-hour exposure

at 670 0C under 100 to 200 mm Hg of CC1..

3. Crystal Surface Evaluation

The object of this study is to establish fabrication and surface

preparation procedures (in particular, optical finishing) which avoid

surface features that limit the material's optical transmission atlO.öfim,

resistance to corrosion by water vapor, and mechanical strength.

The first attempts to cut KC1 made use of a steel wire using

320-grit silicon carbide (SiC). The use of glycerine as a vehicle was

found to be unsuitable. Because of solvent action on KC1, the saw tends

to bind and break. Solvent action varied with exposure time and cutting

rate. The region close to the source of the glycerine-grit suspension,

introduced by gravity, was badly etched. The resulting cut was quite jagged.

25

•'■■'—^ ■

i iiiiwuinnnepSBffvumw^iiijiip^ ,IIHI:«M ■-^.i.n < mapillpitnMllimuiwiiii IMIllWPH^JIPJtWlUfflWWffBWPWilWWBW^»!!!.»1.'!! ^iJ'ü i-i1"" H."iWpWWWWPWW^1||WWWllWI|ijp*MB

1419-2

DEPTH, mm

Fig. 9. Harshaw KC1 Scrubbed (20 Hours, 670oC) with CCI4 Vapor MOO mm Hg). Pene- tration Curve Across'Cleaved Surface (8-mm Thi ck Crystal).

26

-■ ^ :"-- ■ --" --^^--^——— 'tt*^-j-^-1^—— ' '■- - mm "--~-^

(a) KC1 surface as cleaved in (100). (b) A similar surface after 20 hours at 670oC under a low pressure of helium, (c) A similar surface after 20 hours at 670oC under CCl, vapor..

Fig. 10. Behavior of Slip Lines (110) in KC1 With Thermal Treatment (Viewed in Pol'ari zed Li gh t) .

27

■ ■ '' ^. — -:.^^.v.^.~.—^tJ^—.■«.-■■—^.^-t~v- - -■ ■ - ■ rMiliUftrHiVlI. -■■■ ^_ — Ä

II I Mil ^».w-..-—..—.». ^.-—..-« -^«- -,- ,.« . ...MW U fiJJ^IilJIllUlWWWUl-WHMUWJ^

(a) KC1 surface as cleaved in (100). (b) A similar surface after 20 hours at 670oC under a low pressure of helium. (c) A similar surface after 20 hours at 670oC under CC1, vapor..

Fig. 10. Behavior of Slip Lines (110) in KC1 Wi th Thermal Treatment (Vi ewed in Pol'ari zed Li gh t) .

27

..:*.>■-rm^r^z:....-..^^^--.-...- .^^.^~^„ ^ -.-^ —^^—^—^ .

mmmmmmmmmi9 1 """i""" i ■" MM

W

<

Lfl ^ 00

1 0 • r-t 00

ir N 't OJ ^j ß H — ^ X.

3 cc m 4-> re

O O o en »H ■'-' u ,—i

• r-i C' rv^ in rt ^ 0) 0 0

* I N -H o

+1

o

1 -H

o H T3

> 01

^ iri t- 0 +J • rH ß vD 't Q\ U

U tn • 03 0 0) .H T3

1^ fl) t+-) 01

u

t^ 3 a; +J ■ 1 rC

0 c -5 0 (1) a

.H r1

Si c tn O

O a 0 '-

PH

^-< re u

•H c It

rC U

C 0

s £.2 2?

71 ^ > a C

re

g

• 1-*

o c o ft

Lfl

M. vi

VI

{Z ~ — VI

i—i *-.

. c

I--

•> VI in

• "^ cr- vi

0 r-

a; c >- cu c o .

■f-< rt

"1 ;, were

exte

n

tus,

th

e sh

ß •r-< i-H

re o

ß re

OJ 11 ex, 00 r-l t P •r-*

tn ^ rC

s» *> c • -j' Q 3 ^_. ^ ß u

o 2 -♦->

It o 3

T3 - r- H a

o o

0

u

0 u I—1

> 03 OJ

T3 T3 ^-4

m -^ o

nd

IIA

P

ol"

KC

I a o 0 r]

|2 o

Lfl

5 CO

ro — f

-5 ** c in o re a

(U o

• I. it 0 -J ■H +1 r—I *>J.

Clj 0 0 o- ft M C

• H i—1

< • o

in •

o

o 4-» r* r

O ^H

Ü c

re rj

• ß ö0,H

rt '— —. — H n ^ u J

o c o o (H ••

■S.2

■i-> »-1

ß

< o

O

■- u

m Ü d) u

> u

m ui ra 0 3 0

u a

•— W

5 •f-t

rd

4-> n

u

0

u — 0 +

4^ Co

o o

0)

3 in p

are

nth

e

ryst

al

surf

men

t w

as

n ro

lysi

s o

cc

list

akable

.

4J Qj

0) •"

is 0 X. . r-

oU

(/) m ofJ - o -, i «

O -H

71 ^ r ß ^ >> ß

PH 1 1

r- M C "1 c M c s- « t M

c 0 ' "-< ■•*■ »rH -f- .H in e ^

his

me;

o

ub

t th

a IC

l) w

a 13 u a

<s -cU TJU T3U m 3 ,; u >,,"—■ r-1

•iH

u > 0

OJ 0

0) r^

o o Ci 0 60 r- in

0) c 0, _ß 4) 4J

■^ V-, D r-H ^ — sO — ^D H H T3 C H 0

W U U U u

28

■ „^^^^^^^^„.^^^..^.^.^^W^L.^^.,. tttfa ...... , . . , ■ -..g

'«îPtîyiUlimJNPp«.!*!!»! PllMJil1l..J|J».l|lip^WWIIJll|il,ll|WPJiWUJlJ*^lWlll|lt^

A much smoother cut was effected with the use of a nonsolvent

vehicle. The nonsolvent chosen was a heavy hydrocarbon, since

OM-derivative liquids of low volatility (such as glycol or glycerine)

increased the surface absorption in the 3- and 10-[j.m regions. How-

ever, hydrocarbons are not completely inactive in these regions (C-H

at the 3-|j.m and C-C at the lO-fim regions) and should be removed

after cutting. These results are discussed in Appendix II. A com-

parison of the cuts obtained with the two liquid »vehicles is shown in

Fig. 11.

The next comparison study dealt with the combination of abra-

sive and cutting wire. Bridgman KC1 and NaCl ingots were used for

this purpose. The previous procedure, (slurry of 320-grit SiC in

light mineral oil) took 30 to 50 minutes to cut across 3-cm diameter

ingots; the cut was not very uniform. An 0.008 in. wire impregnated

with 45-|j.m diamond, lubricated with the same mineral oil, effected

a uniform cut across in 4 to 7 minutes. The latter apparatus was

also shown to be applicable without the use of the hydrocarbon.

Wo have already considered the effect on the mechanical

strength of the bulk material (KC1) of the method of generating the

surface. The results showed that wire-saw cutting was preferable

to cleaving. The relevance of the annealing procedure following the

primary step of generating the surface has been shown. In addition,

data showing the pertinence of the extent of surface rework and the

nature of the materials used in the last step (optical finishing) to

surface absorption at 10.6 |im is given in Appendix II.

The effect of cleaving and polishing damage on (3 was studied in

three sections cleaved from a Czochralski KC1 boule (sample C-58)

which was grown from scrub-cast JM (99.999%) source powder. The

Convoil-20 from Consolidated Vacuum Corp. of Rochester, New York.

Lastec No. 2005-C from Laser Technology Inc., North Hollywood California .

29

"'■"■'-•-""lii^JifciiM'jllBlliri-1. r'll* mimlim' m J.lî-,'J.-...E...LfÜ,...'....*■

WMii^jMHW» ^■.■7? ■..*f«w,wji'»i(«»w-f»,-i.*i

'234

RESEARCH HUGHES LABORATOfm*

Fig. 11. Wire Saw Cut KC1 Using 320 Grit Silicon Carbide. The Specimen to the Left Made Use of Glycerine as the Vehicle and That on the Right, a Heavy Hydrocarbon.

30

*■->-'■- ■ • — --i u jiiiiiiiiiiiiiiltiiMliililiHiÜiMilliilli " --" ■--' - .-- ■ • -- ■■—

I II I pwma> ■ ■■»«mm» >■• i i i mi i ■ i lum mmm-—*~~~m~~ mmrm^^m i IBII m mi ■"■^^H

three sections were identified as L2 with L = 1.90 cm, L1A with

L = 1.22 cm, and B2 with L ^ 0. 57 cm. Figure 12 shows the batch

between crossed Polaroids. The sequence (a) to (d) is given in

chronological order. Figure 12(a) shows the sections just after they

were cleaved and polished (linde A in alcohol). Figure 12(b) is the

batch in 12(a) after a 20 hour heat treatment; 12(c) is the batch after

an additional 30 hour treatment. Surface etchings in 12(b) and 12(c)

are seen in these unpolished pieces. These surface irregularities

were removed by polishing prior to the measurement of (3. Figure

12(d) shows the polished batch in 12(c) after ß was measured. The

amount of damage reintroduced by polishing is clearly revealed by

12(d). As shown in Table III of Appendix II, ß increases with repeated

surface polishing. In the calorimetric determination of ß, the slip lines may

account for a significant fraction of the laser beam being scattered

directly to the thermocouple. The surfaces shown in Fig. 12 are

parallel to (100) and the slip planes are (110). That the damage occurs

over the bulk is easily appreciated by observing in another < 100>

direction under crossed Polaroids. The slip lines are seen to propa-

gate under light squeezing to a stress value easily achieved in con-

ventional handling in polishing or packaging and mounting for shipping

and storage. The results of the measurements in ß on the specimens, shown

in Fig. 12, are collected in Table III. Within the accuracy of the

measurements, the results for the 20 hour and the 50 hour treatments

for section L1A are the same. In spire of some damage being

reintroduced in polishing. Fig. 12(d), Table III shows that the heat

treatment achieved a factor of two reduction in ß over the value for 0-hour

treatment. The polished section B2 after the 50-hour treatment was

damaged accidentally while positioning the laser beam. A crack

developed at the edge which propagated internally (Fig. 12(d) ).

31

- ^ ■■■■■'■■■—- — - - -

fi'iii ¥ wmwimwiimmm*»* 'i-.\--<i .«'T"-,-.W«T '^wi?»-7wwr!www|^v^!!pifl^(f(p^qw^^

a) Cleaved and polished, no heat treatment, (1.7x)

b) Batch (a) after 20 hour heat trea tment, unpolished, (1 . 7 x)

(c) 'Jatch (b) plus 30 hour heat treatment, unpolished, (2.Ox)

(d) Batch (c) polished, (2.Ox)

Fig. 12. KC1 Boule Sections Seen Under Crossed Polaroids Each Batch Photographed Shows L2, L1A, and B2.

32

■ -- —

ijim.pupji wwrnimMPi.!« WWWW«fWipwr?(W^^CT»fPf»ip^ff^î»flin^w^^wWPW7P^^^

TABLE III

The Value of |5 in KC1 Sections versus Cumulative Time in Heat Treatment

Section Identification cm

Calorimetric Absorption Coefficient ß , cm" '

0 Hour^ 20 Hours' tt 50 Hours"

L2 1. 90 o. nni2 0. 00063

L1A 1. 22 o. nni2 0.00050 0. 0005 5

B2 n. 57 n. nms 0. 0010

"Exposure to inn to znn mm Up of CC14 vapor at 700oC.

T Sections obtained by cl treatment.

âvinf; and then polished. No heat

Sections from f after a 20 hour treatment and polished.

ttSections from - after another 30 hour treatment and polished.

Section accidentally damaged by laser beam.

T594

For the study of surface resistance to water vapor corrosion,

(isothermal case) there are three vapor pressure regions of interest.

Representing; Pn-,0 as the ambient vapor pressure, the regions are

defined with respect to the vapor pressure of the saturated solution,

P{j-,0' ancl the solvent, Pj-IpO- The more common situation is where

ÎIpO - PlI-jO- The stability in this vapor pressure region was studied

with respect to ß. It was shown that hygroscopic behavior was negli-

gible in single crystal KC1 (Appendix II).

As the ambient vapor pressure, Pjô» increases, the inter-

mediate region is reached where PjjjO ' ^HÔ ' ^HJO- ^n this

33

■ .^l.iP»,|»|Pl*'»"WV-«^..:>,-Ull.-rl.l-AJPli"l i--</f%IiW<v-u'?« M"1

region, an unstable crystal will not only mar its surface by hygroscopic

behavior but can cause water to condense from the atmosphere because

^I^O > ^HyOt and, therefore, deliquesce. The extreme region is,

of course, the dewpoint region and above (i. c. , PJI^Q - ^HoO^' Tran-

sient residence of the KC1 single-crystal surface at this extreme region

was studied.

The surfaces were generated by cleaving and subjected to various

treatments to assess their effect on corrosion stability. Fleeting

residence above the dewpoint was achieved by careful breathing on the

surface of the crystal which was observed under a microscope. The

breath effects a dense decoration of the surface with microscopic drop-

lets. In a room held at 25 C and 40% relative humidity, the layer

evaporates off in ^5 sec. After repeating the cycle a few times, the

surface appears unmarred to the unaided eye. Examination was made

on a microscopic, scale of such a response in KC1 after one cycle using

surfaces which have been prepared by cleaving (see Fig. 1 5); cleaving

followed by annealing in He at 670 C for Z0 hours (see Fig. 14); and

cleaving followed by annealing in CC1, at 670 C for 20 hours (see

Fig. 15). These three photo sequences illustrate that the surface

free energy is consistently different for the three methods of prepara-

tion and decreases from Fig. 13 to Fig. 15.

Typical cleavage steps are seen in Fig. 13(a). After one breath

cycle of condensation and evaporation, Fig. 13^b) shows the irregular

outline of the condensate pattern. Etching action by what was once a

droplet occurred at a cleavage step, A higher magnification of the

central region, Fig. 13(c), reveals square pits as large as 4 fj.m to the

edge, with a hopper pattern extending to a depth of 1 to 2 [jim. Such a -12

pit corresponds to the etching out of 10 ' mole. Initial volume ot the

droplet is 10 cm . The volume of the saturated solution is three

orders of magnitude smaller if one considers the fast dissolving region

(i. e. , the etched region) as the main source of solute (total dissolved: -11 - -5 2

10 mole). The maximum area of the etched region (10 cm ) is

34

— --- — -■ -■■ -■ ii in -

r ipimpi II.I™^!!.. <^">PI'<i~niiii.M>u luii! Mill iiiujji.Mii^iiLlUJIII^BVnpBqrmVI^ IM"".'.—»■ J ui iniiiiiu UHU 1 '

K

o «c

O) Cn r— OJ O. U E S_ O -r- U

(U S- x: 0) +J

OJ +J 03

OJ

C OJ O r— U D

O — S- m -a

4-> (O

4- CT. c O) ID C -r- U

•r- C '-^ 3 QJ O X O .c O ^= +J W o in fo c^i 4-3 —' - o

OJ -M QJ i— Qj 'S o u cx: io >,-.- 3 4- u i—

3 4- 3 c CO O O «3

Ol

C i— ■r- — "D oi ^: re CD

r— to i- o D <D fO o^~ o.^: S- X ^

TJ O 10 1 O <D S-

tz O cn OJ •.- r— "O Q. x:^-- cu D. +-> o •^ -o x: x: 2 cu +->

•p- 'r— ro C4- S

o •.- -a •r- C L0 C CTi Oi-M fO —

J^ —- (O--^ 0) ^ O 3 o u +-> c cr o (D

—- LU -r- l/l +J 4-

c

T5

OJ

0)

<-) • >, O) o s-

3

.c +-> +J to fO ■1-

0) o L. 2;

CO 4-

 na u OJ 4- 4- O 4-

LU c O

QJ • r—

x: +-> +-> ro

S-

cn o c: n.

•r- fD g > o UJ

-C oo "D

C

cu fO u c c CD o 3 •1-

cr+J cu ro

on to c

o cu +J -o c c:

-C o Q- o

35

u ■.-.J^.I , ..-.^t.. .. - .i^mtVi-r''ii---" ■'-'■- ,...^^ ...■J.^....

, ■ '■-•""J.".«HWI»,Wl**,tMHW"+l J» w'WWWWiHIJ'pnipWWBWrWPWBpPfp^^^

•

f

i %

:.

fi

_ 4- CD fD O « ■(->

CD ,- 1— CD i— -M OJ re i—

U i— CO r- a. >. c c i- £ o • 1— cu QJ o

Q.+J u ai (U X re ^r jr QJ E 4-) +-> +-> o .

CD X) c c+- s_ M- r~ CD M- ai O +J ■M jr: O +J • r- cr. H- (U +J 1/) 3 "D (O c re o o CD

•r— c .c TD ^-. 1 c OJ +J C

re +J OJ X! 1— 3 — 3 ai 1 re O

O CO s_ s- c o r*

•i— S- (D Q- l/l CD re ^D ro c S-

c •— > o re o 3 C ■1—

•r" cr ro ai cr r cr GJ U c CD X) QJ S- •i— S- CD S_ S- a; 4- >

t-H 4J S- i—

OJ re >> CD o E • CO v >> l/l re ^-~. C CD CD on

00 X OJ > r— •r- o XJ ■o a c CD 0)

-C CM o -E J^ >> u +J t- 1—

o c

CD

3 o x: C/l

X o o o

S- CD

-M c CD • U L0

CD ••- x: a.

CD

O +->

cr. c

•I—

XI re CD

CD

o • >, CD o s-

+-> CO re T- CD o

ca c*-

CD o

O E

o CD H- f- O

UJ c: o

CD ■.- ^ 4-> -i-> re

i- C7) O C Q. •i- re

3 > O UJ

00 X3 c

ai re o 0) o 3 •!- CT+J <D re

oo to

o 0) +J x> O E x: o a. o

Dl

36

kkk. . .- i illillitltl . ^.=.. 1. .-..-^.^w-^- . --■■■ . , , ^_ ^

im i ■■ i mmmiu i ..iMi.np» ,,* w*mm^mmmtmm^mv" m 11 ■■n i '" ■ ■' ' ■ l i" 11 ..... . .

i f

r

D ^

\ T

■M w 0) c w J-

C— •!- 3 0) o 'O c >, £= s- u 3 - o

o c u cu s: ai ^: to 0) ■.- o to

+-> 1— C (U M- +-> ai IT3 3 co "O

O •^- O)

m i- ^^ OJ fO J3 ^ •

r-~ —.4->^-^ C 3 ^ <0

•r- Ol r- 4-v_, 0) O

c s- -o c o t. C »♦- -r-

•r- •■— rtJ <+- cn O lO QJ QJ .—^ Q. S- -C xi cn Qj

h- -— c +J OJ ro -r- 01 E • r- -o (O^—. c ai oo x • 3 .c

o W O 4J -^o ai s- xi oo •1- If-

-U. fO o

• o 1 >> QJ O 0)

£1 c <— -o IS. o-^

■a (a s- cn S- o; 3 >! o s- >, o OJ J3 Q. ra J3 ^ &- ra a. ra "O > . Q- S Q- C -Q

o s- QJ 3 ^-x— O -t-> o c X i— M- 10 J3 o O o •r- O 4- C_) QJ to 4-> CM O CD Q. ra -—-u O ra QJ s-

c r^ > 4-> o a) re ID ra to o o QJ QJ (O^-^+J -— 3 "C •

M- o ra O QJ ,— s- o E >. ra 3 !— +J . +-> •r— l^v_^ C —> to c s-

0) cn 3 ra QJ >— +-> E ^— o u +-> ra rrj +J ^■-E U1 ra

•r- ra i— to E 4-) O) OJ o >> ■r- C S- O QJ s- -a C -f- -t-) o QJ QJ

1—1 > S- ra > ■(-)

ra 4J QJ 4- .,_ ^^. OJ ra -a s- -a to ra r— 

X QJ O O "O 3 O E 5- O 3 ■•-> i— O tO

&- +J «4--— ra QJ o u

.E ^^

J3 T- r— — s- o

O > "O *+- M- QJ E O "O ra

-t-> S- Q. CO > QJ QJ .r- U

QJ

rc QJ

D- cC OL

C o

to 4-> O QJ

M- <+- UJ •

to S- QJ o U Q. ra ra 4- > s-

3 s- oo QJ -M ! ra O 3 ^

37

^._.^..... ..• w^, - —." ' "' ^■-^..—■■

'i i«r.VAWffi£p:Wliq MWUinitvUHLui ■. i' -.™^ vw^v.«••■■ .-^..w^W '^ ^"^T't. 11M. ^Hi-HfM i.^JUHWIMW«! ■ "*■ "^m I>"TWW;WW IMII, fl^iJ-.» .".w.i F L ■I,(. V«>I i up JII ■uijiup ii,L|ijpiuiiW J UW, I ..I li .(.»PffWi ■ uJ ,i.Wk'.«.i!W*J wjpjw.iupi RU.qnin.'PUmiiJI W.»1^™J.1J»(LI.Iwmw«j»BllHA,lfi^

assumed to correspond to the crossover from the undersaturated to the

saturated condition. Redeposition of solute would account for an aver-

age depth of < lO-1 [am. Hence, under the present resolution, only the effect of etching can be seen.

In Fig. 14(a), the cleavage steps are almost obliterated. Coher-

ing to the substrate are isolated regions of growth that were vapor

deposited during the slow cooldown of the thermal treatment. The effect

of one breath cycle is shown in Fig. 14(b). The small new regions of

growth have been removed (dissolved) and the sharp edges rounded off.

The irregular outline of the condensate pattern is seen. Magnification

of the central portion, Fig. 14(c), shows no etch pits.

In Fig. 15(a), the cleavage step? are gone, the s irface

rearranged to give prominent steps bounded by (100). These steps

accommodated growth from the vapor phase during cooldown, as evi-

denced^by the low substrate density of isolated new regions of growth.

The effect of one breath cycle. Fig. 15(b), merely smooths out the

sharp edges; no irregular outline of a condensate pattern is seen. High

magnification of the central portion, Fig. 15(c). shows no etch pits.

The interface of small crystals formed by recrystallization at

the surface has been shown to be a source of initial microcracks which

cause premature failure (Ref. 8). As seen in Table II, the surface

conditions of Figs. 13(a) and 14(a) correspond to a Knoop hardness of

10 kg mm" , while Fig. 15(a) yields 12 kg mm-2. In the case of the

modulus of rupture. S = 640 psi for Fig. 13(a), S = 450 psi for N

Fig. 14(a), and S = 970 psi for Fig. 15(a).

Wetting and evaporation behaviors observed through the breath

cycle provide additional evidence of the stability of Fig. 15(a). The

free energy per unit area at the solid-gas interface is lower than that

at the solid-liquid. The microscopic dewdrops do not have a tendency

to spread (i. e. , no coalescence). According to the Kelvin equation,

these drops are unstable even at 100% humidity; hence, a gradual fading away of the film.

38

. .. -^■■..-^.t^..^--^- - -> -^ ■■ .. -.■■_.■. ^ .> -^ .- -. J-i.^.*-^^ - ■—i-" -^ J.-uw-auAvi^.. ■^a-t^-..iaM^JLi-ri-i>..^|it-rtt r.f . ,^-^^_v ^...t.. iji nui i j| flJlTltM^r^V-U- , , i _.

m mt^mm^rw^^i^^' «■■■«■«■»« i i ■■ i ■• nmmmmimmmmmmimm I I • I II I III! >■! LII HI 111! W^«WWIWIP«W ■ ■■nil uuHii iiiuiii »i

The opposite inequality holds for Figs. 13 and 14. Coalescence

of the small droplets leads to larger droplets or irregular contours as

suggested by the irregular outlines left. Solvent action and evaporation

are mutually governing factors, but both lead to the attainment of drop-

let saturation. In the fraction of the residence time (~5 sec) that the

solution is under-saturated, its etching action depends upon solubility

rate which, in turn, depends on the free energy of the locale. In

Figs. 13(b) and 13(c), etching action is evident only on the stressed or

high free-energy (defect) region. No etch pits are seen in Figs. 14(b)

and 14(c), presumably due to the thermal treatment under lie which

partially relieved the stressed condition brought about by cleaving.

The above experiments indicate that considerable Su>face

stabilization followed from the RAP treatment, perhaps to a stability

approaching that of a natural crystal face. In the last quarter of the study, Polytran specimens were

received for surface evaluation. Photographs were taken of an as-

received Polytran disc, identified on the Lucite holder as «77. Photo-

micrographs at 83x and 1 66x show polishing scratches in the as-

received surface. Photomacrographs at 2x show the grain si^to be

10 to 40 mm2. The grains developed readily by exposure at 23 C and

40 percent relative humidity. Photomacrographs taken at 2x between

crossed polarizers show a correlation between the grains and the strain

pattern of slip lines. This comparison is seen in Fig. 16. Sections

were sliced from the disc. A piece was kept for control and another

was scrubbed in CCl4(g) at 670OC for 20 hours. Figure 17(a) shows

the control and the scrubbed pieces at 2x magnification. Figure 17(b)

is an 83x magnification showing the surface texture across the grains

before scrub and Fig. 17(c), also at 83x magnification, is after scrub.

In the latter case very prominent (100) surfaces developed in each gram.

39

1 "flW»PWMPIIIHH*PWIWUilliu <>i<.>>i.. <ii"iu<iNiK.iBiiii.(iKunnpppH HI » M,IMiWM(^^I.U«II.IMII« II »I U W^P^^^^^W^»»»» ■"!' •■"•« »ipiWPW«Wl«M^,«Wi ■ i 1 "7- -t- A-»Itl KIL». ^l..«"L".*'.'-tl!*"—»B?TW1^(

(a) Polytran disk No. 77 as received (2x magni fi cati on)

(b) Same as (a) but viewed in polarized light (2x magnification)

Fig. 16. Surface of Polytran Disk No. 77,

40

.■-■ ■ ■- - -- .—...■.^^.^..^^..-^■.i-^....-.^.^^^- — ... - ■—^—-..— -.. ,...,...,........,..-.-. -- -,.-

JJ|Ji.U|l||lll|JJJBJ»PH"«.«,W!"^,"l"»-."iWJ.Wl'^'< «^^nnpin^ffpi*<w*«<wp^«ipwninvvn^^

Scrubbed Pi e ce

Cont ro1 Piece

(a) Polytran sections (2x magnification)

(b) Control piece (83x magni fi cati on) (c) Scrubbed piece

(83x magni fi cation)

Fig. 17. Polytran Surface Before and After Scrub.

41

ii Mm i n ■imrBiiMni - - - . - —. I, ■ i i ■ ■

mmmm. nmM

4. Purification and Analysis

I

Powdered KC 1 five-nines pure in the metal ion is easily available.

However, the anion purity may not be any better than three-nines pure.

Before this study began, we had demonstrated that various anions;

e. g. , NO" HCO" CO^,, C?0~ etc. , of the alkali metals can be con-

verted quantitatively to the chloride by treatment with CC1 , (Ref. 9, 10).

It was also known that the chlorides of many heavy metals are quite

volatile. It was, therefore, possible that CC1. vapor could be used to

effect simultaneous cation and anion purifications of molten KC1.

With purification as the objective, the casting and zone refining

of KC1 ingots under RAP, He + CC1 ., was developed as shown in

Eig. 18. The analyses of the starting powder and the RAP treated

specimens are shown in Table IV. Emission spectrograph analysis

turned out to be inadequate to differentiate between the two powders,

MCB (99. 9%) and JM (99. 999%). The method was also unable to dis-

tinguish the purity level of the MCB powder from the various sections

of the zone refined ingot derived from it. However, beyond a doubt,

the volatilized material, or scrub deposit (sixth column. Table IV),

showed that various impurities had been transported through the vapor

phase.

The study was complicated by the fact that the boat (vitreous

carbon) used in zone refining was a source of impurities. Table V

shows that under CC1, vapor, Ca and Mg, unlike Si and Cu, have a

high escaping tendency (fugacity) from the vitreous carbon matrix. The

fugacity across the solid-vapor interface is a measure of leachability

by the halide melt. While Poco graphite is more suitable in purity,

it is inapplicable because it is fairly permeable to the melt.

Zone refining conditions; The moltun zone was ~2 in. wide and was moved at the rate of 5 in. /hr. Three unidirectional passes were taken.

42

i • r ! r uniiViriMri ■'-■ ■ --■"- ---' 't-*-- ^■—.- ,. iimfiftüimi

fm in" ■ ■■ ' '■ ■ ■ mi^m^rm"^~m*~m ■ mm in ii .«■■■■ mimutwmmimmwmiim ^m^m^mmu ii i ■ i ■■■

(a) Scrub-cast ingot, 3 x 14 cm

(b) Scrub-zone refining

Fig. 18. Alkali Metal Chloride (KC1) Technology

43

■ ...„.-w, J,- M|,|- .M. I' - ■ .- ,-^.^.^.J.:-^- - -^.■-■■.^—.,

r ""■ ■ "°—^•'•^>ii ■! ■! i ■ ■ wiaHpn^OTMM^n' -^■«•PWIi

TABLE IV

Emission Spectrograph Analysis of KC1 Powders and Crystal Specimens

1 Zone Refined Ingot MCB Powder

Scrub Depos i t

JH Powder

Nose Center Tail

K 52% 52* 52J 52t 50% 52« .

Si TR < 0.002 t

ND < 0.002 0.0031 NO < 0.002 1.5 ND < 0.002

Ca 0.0013 0.0012 0.01 1 0.0012 0.040 0.00045

Fe NO < 0.001 NO < 0.001 ND < 0.001 ND < 0.001 0.15 ND • 0.001

B NO < 0.004 ND < 0.004 ND ^ 0.00 4 ND < 0.004 0.019 ND < 0.004

Mg 0.00014 0.00010 0.0017 0.00032 0.0039 0.000085

Pb ND < 0.005 ND < 0.00 5 ND < 0.005 ND < 0.005 0.026 ND < 0.005

Cr NO < 0.0001 ND ■- 0.0001 0.00019 ND ' 0.0001 0.017 ND < 0 .0001

Bi ND < 0.001 ND < 0.001 ND < 0.001 ND < 0.001 0.0067 ND < 0.001

Al HO < 0 .0004 NO ■• 0 .0004 ND < 0.0004 ND < 0.0004 0.015 ND < 0.0004

Cu ND < 0.00005 NO < 0.00005 0.000074 ND < 0.00005 0.0024 ND ' 0.00005

Li ND • 0.001 ND < 0.001 ND < 0.001 ND < 0.001 0.0031 ND < 0.001

Aq ND < 0.00007 ND < 0.00007 ND < 0.00007 ND ' 0.00007 TR < 0.00007 ND < 0.00007

Na ND < 0.005 TR < 0.005 0.015 TR < 0.005 0.057 ND < 0.005

Ti ND < 0.001 ND < 0.001 ND < 0.001 ND < 0.001 0.0019 ND < 0.001

Ni ND < 0.0005 ND < 0.0005 ND « 0 .0005 ND < 0.0005 0.0015 ND < 0.0005

T TOO

44

■ —■—— - ■ ■

rTrr-,.|.iji.nj.B(Wi»i»_p»nLii ^...mvHm^xm^wwvr, ,Tvwwm**wvi^m^**wm*^\^^wmm\>*\i*mu\Mmmmyw

TABLE V

Impurity Content of Vitreous Carbon and Poco Graphite

Impurity

Vitreous Carbon Poco Graphite

As Received After Scrub As Received After ZR "

% %

Ca

MK

Si

Cu

Other Elements

0. 0026

0. 00086

0. 0048

0. 00026

Nil

0.0016 TR<0. 0002

ND<0. 0003 0. 0023

0.0041 j ND<0.001

0.00032 0.00021

Nil j Nil

0. 00065

0. 00060

0. 060

0.00075

Nil

Exposed to CC14 vapor at 700 C for 21 hours. : Exposed to three /-one refining passes on molten KC1 (~800 Cunder CC1. vapor) for 21 hours.

T593

Impurity cleanup in molten KC1 under CC1 . was demonstrated

directly by employing concentration levels of the impurities above the

sensitivity limits of the emission spectrograph. The results are shown

in Table VI. The vapor pressure of molten KC1 is ~1 mm. An approx-

imate measure of the order of the relative fugacity of the dopants is

the temperature of the pure form at its vapor pressure of 1 mm:

FeCl3, 1940C; PbCl2, 5470C; NiCl2, 6710C; and MgCl2> 7780C. To

correlate the order to the relative cleanup achieved in molten KC1,

the evaporation flux of the solvent needs to be determined.

45

■ -. ..■,-.^. - ...^ -^ - ■: ^.^.^-.-,^^^. ^ -... . ^: yi ,,..,.^._u, if[iMr*ill

„,1WV,1T,,.-V.„.. -,11 -W'l.' "■■W-«l ,IIII»1. MII^(i,|^»Hl|l HWti^VKf^-TO—

TABLE VI

Impurity Cleanup in Molten KC1 Under CC1 .

Impurity

Exposure Time of Melt to CC1.

0 Hour"" 5 Hours 20 Hours

Weight Percentage

Mg

Pb

Ni

Fe

0. 0050

0. 22

0.032

0.023

0, 0016

0. 039

0. 020

0. 0010

0.0014

0. 041

0.035

ND < 0.001

Averii cd over the ingot obtained from melting under He and freezii he melt rapidly.

T592

Consider first that, for a given impurity, a conservation

relation holds,

r M) (i)

where x. is the initial concentration in the melt (weight M), x is the i s

concentration in the sublimate (weight m), and x^. is the final concen-

tration averaged over the residue (weight M-m). In terms of measur-

able parameters.

m ,. M ^ ft<r (2)

46

■ ■ - -■ -'— ■ *■' -- - J..J,,.,->...-,.-,j..„.^„.....

-■iaiBI,^K.îlU>-V**T

-2 -1 where f is the evaporation flux (g cm hr ) at constant temperature

(T > T ), t is the soak duration (hr), and cr is the specific evapora- mp 2-1

tion surface of the melt (cm g ). For the process to he practical,

the first constraint is

rn M << 1 (3)

i. e. , the evaporation loss of the host material should be low.

Two runs were carried out with KC1 in a quartz boat held at

840OC, at a gas (He + CC14) flow rate of 2. 5 x 103 cm3 hr"1 and a

mole ratio of He:CCl . = 7. The results are as follows (a = evaporation

surface):

Run 1: M = 7. 36 g, a = 10. 3 cm , t = 5. 0 hr, m - 0. 169 g

Run 2: M = 7. 72 g, a = 11. 2 cm2, t = 20, 0 hr, m = 0, 560 g

The value of m was obtained by the difference in weight before and

after soak. The average of the two runs yields cr = 1.43 ± 0. 04 cm g -3-2-1 and f = (2. 9 ± 0, 6) x 1 0 g cm hr . For these values of f and a ,

an overnight scrub (t = 15 hrl is a practical processing time because

m/M = 0. 062 according to eq. (2), which meets the constraint shown

in inequality (3).

The second constraint is seen in eq. ( 1 ) in terms of x /x., the si

relative volatility factor. Impurities with

s_ > M _ m (4)

In ideal-solution behavior where the host (solvent) follows Raoult's law, P] :: (1 - xj) P^, and consequently, the impurity (solute) obeys Henry's law, P2 = ^Pf • ^ can be shown that xs/x. i P^/P? when (x., x ) << 1. '" l

1 s

47

^-.û^.^.^.-.. ^J.„â.^.jks^.^J^JJ-^^^.^..^,..Jlä|iM ■ „.—....^■-.^._. ._ _...._....... ,... - - -

I I'll I II i mmmmm^~* i n n u ~—^Kmmm*^^*mimmmmmmamm "■^i

would have been stripped out of the melt at t < 15 hr. Only those with

M m :5)

would still i)e detectable in the residue (i.e. , x measurable). For the

given case, M/m 16. A relative volatility factor of 10 would strip

out only 60% of the impurity (i. e. , x /x. = 0. 40).

For the impurity cleanup work reported in Table VI, 2-1

o- ^ 0. T cm g . From the value of f, it follows that m/M = 0. 007 3,

satisfying constraint (1). Using the 5 and 0 hour values to calculate

x^/x. , where x is the final concentration ave raged over the residue,

the xsm/x M value for each impurity in decreasing order of volatility

from the melt is as follows: Fe , 1.0; Pb , 0.8; Mg , 0.7; and

Ni , 0. 4. The actual volatility order shows that NiCl has a relatively

low fugacity in the melt (KC1).

In the /.one refining experiment whose analysis is reported in 2 ? l

Table IV, M 450 g and a 170 cm ; therefore, a = 0. 3« cm g .

The molten zone had a residence time of 0. 4 hr. With three passes,

t - 1. 2 hr. It follows from cq. (2) that M/m = 760, It Is seen that Si

has x /x > 750 and satisfies eq. (4); i.e., (x /x.) • (m/M) > 1. To si si this class belongs the volatile chlorides of B, Al, Fe, Cr, Hi and Ti.

The latter group, taken collectively, is characterized by x /x. > 200. s i

The others belong to the class characterized by eq. (5), i. e. ,

'Xs^Xil ' 'm/'Ml ': ^ Because of leaching of the vitreous carbon boat

(Table V), Ca, Mg and Cu account for x /x. 30, while the rest amount s i

to xg/xj > 8« Using the impurity-class content of the original powder

as the weighting factor, the average for three zone passes under RAP

is x /x. > 100. The purification effected is even better if we include

the effect of segregation in the condensed phase. The results of these

calculations lead us to believe that RAP purification is capable of

converting KC 1 from 99. 9> to 99. 9997« purity in both metal ion and

anion species.

48

._M>>_M-l_BaMaM_aMMaA_>a_aaakMMMlMi mmtimjmmiimmm^ MMIIHaMHa-HaMM »■■■■feBflliMMMB

L^M»n>i«mjH|f^p|np^MHHWU9flfPIS<nn9*nmnin|Vl'VJ^^

Our initial emphasis was on the development of a procedure for

purification; it was convenient to begin the study by eliminating OH~

as completely as possible and comparing the resulting physical prop-

erties of ordinary material containing an unknown quantity of OH". Of

course, the most convincing evidence that OH (s) is the culprit would

be a determination of the physical properties as functions of directly

measured [OH (s)]. In keeping with this objective, the last quarter

was spent comparing analytical methods, destructive and nondestruc-

tive for the quantitative determination of [OH (s)] in KC1 crystals —

acidimetry appears to be the most reliable. •7

As pointed out in Appendix I, one needs to attain PH?O/PHC1 - ^~

in RAP to suppress hydrolysis. At the melting point, the equilibrium 2 - 5 constant ol hydrolysis may be as large as 10 . Consequently, C S 10

is tlic requirement on the sensitivity of applicable analytical methods.

In the nondestructive category, there are two direct spectro-

photometric methods, one indirect method based on color center

formation, and a method based on the measurement of ionic con-

ductivily. * The latter two have been utilized most extensively as

analytical methods to assess general purity of the crystal.

In the conductivity method, the plot of the logarithm of con-

ductivity against reciprocal temperature yields a knee temperature

which divides the region into two modes of conduction, the extrinsic

and intrinsic (high-temperature region). The position of the knee

temperature is sensitive to both [OH'(s)] and total heterovalent cation

impurities. The inapplicability of the method for determining

[OH (si] lies directly in chis feature of nonspecificity.

The potential usefulness of the method based on color center

formation is demonstrated in the observation that X-irradiation

decreases the uv absorption of OH (s) as well as the IR absorption of

the free substitutionai OH (si and the narrow line spectrum of the

The use of (si in Oil (si specifies the phase, i.e., solid, and the 1)racket refers to the concentration.

49

^-^o^

,..i|l.u».Mli»l.inUW.M^ ^-"^^^^^m^^^^BRnW^TOWWT^^-WT^^^TOr-^TTSWJI^^

complexes (Ref. 6). Further, it has been shown that the F band

produced in NaCl and KC1 by ionizing radiation is proportional to the 7

initial OH (s) absorption band. However, such colorability is also 2 + dependent on background impurities; e. g. , great enhancement by Ca

or Sr (Ref. 12). In the case of the impurity Pb in KC1 crystal,

enhancement is observed in early-stage coloration (gamma irradiation)

and the reverse in the late-stage; i. e. , a dependence on the energy

absorbed by the crystal from the gamma-ray field. For the direct

determination of [OH (s)], the drawbacks of the method are obvious.

The two useful regions of the spectrum are: The IR region at

2. 75 pm for the O-H vibration and the uv region for its electronic

excitation, 0. 185 pm for NaCl ana .. 205 pm for KC1. The absorption -21 2 cross-sections are: o"(NaCl) < 3. 4 x 10 cm and a(KCl) = 2. 8 x

10"20 cm2 for the IR; a(NaCl) = 4. 0 x 10"17 cm2 and o-(KCl) = 2. 5 x -172 14 -5

10 cm for the uv. In the case of KC1, it follows that C . ~ 10 Q mm

for the IR and ~10 for the uv. These limits may be pushed one more

order of magnitude lower by improved sensitivity in detection (D . ) o / r / mm and the use of thicker samples (L). Therefore, these two methods are

capable of providing enough sensitivity.

However, results are reported in the literature which show the

difficulty of rendering these metho'ds quantitative for [OH (s)] at _5

C S 10 because of a nonspecificity resulting from the presence of

heterovalent cation impurities. The experiments with KC1 have shown

two classes of OH (s) in alkali halide lattices (Ref. 4). With certain

bivalent impurities present in the crystal, OH (s) reacts with the

impurity metal ions to form OH (s)-impurity complexes. Not until all

of the appropriate complexes have been formed will OH (s) incorporate

It is shown in Appendix III that for a sample thickness of L = 1 cm and a minimum optical density of Dmin = 10"S the minimum mole fraction of OH"(si is C KC1.

rmn 4. 48 x lO-25/o- for NaCl and 6. 24 x 40-25/o- for

50

- - -- -■■-- _ . — 1M— i —■

"wi"" i n.ijivnmpnnpppnnnn^ni^nnn^vvpmmiiiMi »»-«^wniBniiMiii n. IJ ii\i^*mi^m^m^m~^mmmmmmKmim^r*tm

into the lattice as the free substitutional ion. From 2. 6 to 2. 9 |xm, a

sharp line spectrum is produced by the impurity complexes and a broad

line from free substitutional OH'(s). These features have also been

shown in LiF crystal containing the impurities Mg and OH (Ref. 5)

and also in OH' doped NaF crystal (Ref. 6). For these reasons, we

require more than one a value in the IR region (Ref. 14) for total - 5 OH (s) at C < 10 when the total mole fraction ol heterovalent - 5

impurities is ~1 0

In the case of uv absorption of OH (s), it is found that SiO., in

the molten haluie and Ca or Sr in the cryücal remove the absorp-

tion at 0. 185 |im in NaCl pad 0. 204 |j.m in KC1 (Ref. 11). Hence, the

method, like the IR method, suffers similar drawbacks for quantitative -5

application at C < 10

In summary, the results in the literature indicate that the above

nondestructive analytical methods cannot be relied upon for the quanti- -5 tative determination of OH (s) at C S 10 due to interactions with

heterovalent impurities.

We do not know whether the impurity-concentration inequalities

were satisfied in determining the tr values in the IR and uv (Ref. 14).

We assume that they were determined at [OH'(s)] much larger than

the total concentration of interfering impurity ions. However, these _5

a values cannot be applied at C < 10 unless the total mole fraction _ 5

of interfering ions is << 10 , which is not generally the case.

The next question which must be raised concerns the primary

method, i.e., the procedure used to establish [OH (s)] independently

for the calculation of (r from the optical density measured at the given

peak wavelength of the absorption band. This issue was not taken up in

sufficient detail in reference 14. The issue needs to be resolved as

the tr values are used by various workers to calculate [OH (s)]. For - 8

instance, an [OH"(s)] < 10" in KC1, using the absorption at 0. 204 \im,

has been reported. A study of reference 12 indicates possible con-

tradictions because, assuming that cr is correct, the value of [OH (s)]

is meaningful only if it can be shown that the total mole fraction of

51

--■' --

;•'WWIWJTW.?^"'-»"*■ ■ ii.j m .1L..4.. Ji .1.1.1■.■■MIJI I IW.IUI11MIRPpJ«NH.I|>WHIIU.l WJWWWIM*W'-ii'i'"i "!" I ■■,.'IM 11» ,»iilii:iJ J| HI". il>,,i ■«■ 11...,,■» ■>■,*■WK»Tfll,fTH-?''-,''f»«' FU^WJUg|IWIMM«WII W^'fl*'"«^ W"P:".T"' '-1■"■:*■'..1-fWiV«"■*-■ •.'i.''. ^r.'^'W^-f.V"«1!»."«.-."»!»»*«! .Pi,.^.j^^^wnp,JMi•WM>«p.^{5ginqH|l

interfering impurities is << 10 , a limit beyond the present analytical

methods in use. In the case of Group IIA, alkaline-earth impurities,

the detection limit for Ca corresponds to a mole fraction of ~10

with Group IIB, for Zn or Cd , ~10 ; with 3d transition ions, say

Fel\ ~10"7; etc., (refer to Table I, Ref. 12).

The popular resort of employing the initial value of Oil doping

of the melt in place of a primary method of determining [Oil (s)] is

unsatisfactory. Usually, the concentration range employed in the - 3 - 4 doping is between 10 and 10 mole fraction. The practice assumes

that the melt-crystal interface has a distribution coefficient close to

unity and that there are no other sinks for OH in the melt-vapor and

melt-crucible interfaces — the assumptions are open to question. Com-

paring the vapor pressures over the bulk material, say at the melting

point of KC1, it is easily shown that KOH is about an order of magnitude

more volatile. At the given doping range, the melt is very reactive.

Platinum, carbon, and silica crucibles are subject to chemical attack.

Indeed, these corrosion reactions serve as qualitative indicators for

the melt concentration of OH ; the transition from wetting to nonwetting - 5

being observed at < 10

The failure of nondestructive methods to yield a reliable and

practical procedure to measure [OH (s)] led to a feasibility study of

converting a primary method (acidimetry) into a practical procedure; -5

i. e. , capable of operating at C < 10 . Unfortunately, this meant the

use of a destructive method, acidimetry.

Two solution methods are available: pH measurement and

absorption spectrophotometry in solution based on the acid-base equilib-

rium of a dye (indicator). In terms of sensitivity, it is shown in

Appendix IV that a pH-shift measurement precision of ±0. 01 is satis- -7

factory for the measurement of C down to 1 0 . It is also shown

In reference 9, the authors employed mullite (alumina-silica) crucibles which are known to be corroded by molten alkali hydroxides.

52

■■■- - - ■ ■......- -^ -■.. ■■■ "--rii.Mii.ni

■ immmmHmmmmi •• • <• « • """"ill* mmmimi*mmmim^^^^'^^**<'*****'mmimmrmmam i ' < ii in i ii

(refer to Appendix IV) that operating parameters in acid-base dye

colorimetry can yield an absorbance-shift measurement with a pre-

cision matching the requirements for pH-shift to measure C down to -7

10 . Hence, both methods are sensitive enough, and should be appli- -7 -5 cable to the range, 10 < C < 10

An earlier study, on the alkalinity of NaCl crystals, was carried

out by D. Otterson using p-nitrophenol (p-NP) as the dye indicator

(Ref. 6 ). He measured the change in absorbance at 0. 41 5 [im of the _3

dye at a concentration, c = 1. 5 x 10 M (M = molar), and electrolyte

(NaCl) concentration of 0. 3 M. A calibration curve was given for the

change of absorbance in terms of NaOH added to the dye solution and a

correction curve for the dependence of absorbance on salt concentra-

tion, i. e. , ionic strength.

Our spectrophotometric characterization of p-NP solution gave

very different results, indicating the necessity of modifying the pro-

cedure. Using pH buffered solutions, our measurements which are

seen in Fig. 19 show two absorption bands which peak at X, = 0, 319 |J.m

and \ ? = 0. 402 fjim. The X ?-band grows at the expense of the X . -band

with increasing pH, showing an isosbestic ^t 0. 348 fj.m. At [H ] = K

(or pH = pK), where K is the ionization constant of the dye, equal

amounts of the absorbing species exist. Then, the ratio of the absorp-

tion peaks equals that of the extinction coefficients. We have deter-

mined the values of the extinction coefficients and as seen below the

ratio is 1.94. Consequently, from pK = pH, Fig. 19 predicts pK ~ 7.Z

in agreement with the literature.

The powder p-NP was the indicator grade of Eastman.

At the isosbestic wavelength, the absorbance is constant and independent of the concentration displacement of equilibrium between the two absorbing species.

53

^ ^~*'~~-*~~- ■ 1--*****-.*-^ -^ I^.J. ^.-^ -.. - ^.

r ■ ■^.^««■«•■■.•■■PPKva v.run* '■^- ".n -^■-*^":—T:-^.-'. uiju. 11- .i| « I^MI.U. j^ppwnp^ppp^qit^^pp^l^nn^^^llil^^^u.UiuiafffR^iiiii.a M. Mn"V|l"*MPHJHiil.~i MJi . |iP,*iW-Hi^ U "- ■... ■■ .)WilPI^H(?WWr! IWW«. i,. .1 ■niinii:

0.8

0.7

0.6

0.5

ABSORBANCE 0.4

0.3

0.2

0.1

.'1/4 1

C = 4.6X 10'6M IN p-NP

L = 10.0 cm

pH7.2

0.300 0.340 0.380 0.420 0.460

Fig. 19. Optical Absorption of p-Ni tropheno 1 Dye in Buffered Aqueous Solutions. Note the Isosbestic at 0.348 ym.

54

—■.i^*. UJ»*-*»**^«. iai r..n ii'irt' riUinlriiVihi ..

In the ionization of the dye,

11.5/ ^ II + &~ , (6)

the \\<<f species gives rise to the \ .-band and.^~ to the \ -band, 15 From a pK = 7. 1, the value for zero ionic strength, we can show

that Otterson's choice of operating parameters is far from optimum, 1 /2 It lollows from eq. (6) that the degree of dissociation is o ä (K/C)

V 1 '\ and from K 10 ' and c 1,5x10 M, the dissociation corresponds

to o - 0, 007 3, As shown in Appendix IV, the optimum operating point

is O' : 1/2. I>Iote>also that [II ] a v/Kc corresponds to the indicator

solution having pH = 5, 0. Figure 19 shows there will be very small

absorption at X = 0,402 |j.m, A further loss in sensitivity is exacted

by choosing to measure the absorbance at 0,415 |j.m.

In working with p-NP, we, first carried out Beer's law measure-

ments (refer to eq. (11), Appendix IV), The concentration range of -6 -5

p-NP solutions was 2, 32 x 10 M < c < l,86x 10 M, The optical

cell had L * 10, 0 cm. The cell was filled and capped with solutions

mixed in 50,0 cm volumetric flasks. With the solution 1, 00 M in

KCI and 0, 010 M in IIC1, there was essentially no absorption at the 4-11

X^-band and t ; 0,9« x 10 M cm , Similarly, using 1,00 M in

KCI and 0. 010 M in KOII, e , 1. 92 x 104 M"1 cm"1.

For comparing the alkalinity of KCI crystals, interaction with

the atmospheric constituents (especially CO ) had to be avoided.

Unfortunately, cells with L - I. 00 cm were the only ones available at

the time which could be used for vacuum processing. The stock of

water employed for all these measurements was cleioni/,ed water dis-

tilled twice in vacuum over KMnO,, A constant amount of p-NP was

introduced into the cell, using a CH.OH-solution. The solvent was

evaporated. Then a weighed amount of KCI crystal was introduced into

A thicker cell would have been preferable as this allows lower values of c.

55

"k-'-—• '"''-•*■ ------ . .^ —;'-- '■ - —-■-—

the cell. Grinding the crystal was avoided as this operation has been

reported to increase the alkalinity. The cell was provided with a _5

stopcock which allowed evacuation to ~10 -mm Hg, after which water

is distilled up to a fiducial mark in the cell. The stopcock was then

closed, the contents shaken for thorough mixing, and the spectra

recorded with a Gary double-beam spectrophotometer. Water in a

matched cell was used as the reference. The weight of the solution

was obtained and this yielded the concentrations of p-NP and KC1 with

a reproducibility to within ±5%.

According to the mass-action expression for eq. (6), the mea-

surement of the concentration ratio, [H.^]/[./"], yields [H J/K. The

latter quantity can be obtained from the measurement of the absorban-

cies in \, and X . If only one band is available, the value [H ]/K

can still be calculated because of the conservation relation,

[H^] + [^ ] = c. Representing the absorbance by y.

rH+i £2 yl k T^ ■' = ~— ' — » (refer to eq. (14), Appendix IV) (7)

for the two-band method, and

K

£2 Lc

1 , (refer to eq. (15), Appendix IV) (8)

or, a similar expression making use of the X -band.

Our first attempt to compare Harshaw KG1 with RAP-grown _5

KC1 was oased on 1. 00 M KC1 and c = 8. 3 x 10 M in p-NP. Because

of the high concentration of KC1, the one-band method at X.^, was used.

Two measurements with Harshaw KG1 crystal gave.

-U|--J = 15. 2 ± 0.3 ,

56

. . . _ . . : . . ■■ . ■ . ■ ...... A ...-.■

■ n mmunHimmmmi

and one measurement on IIRL B-2 9 KC1 crystal yielded,

[H1] - 18. 5 K

All we can say at this point is that IIRL B 29 is 3. 3 K units more acidic.

At the concentration of KC1 employed, only one absorption band domin-

ated (X. -band); hence, K could not be determined.

The two-band method was employed using more dilute KC1 - 5

solutions. The p-NP concentration was c = 12.0 x 10 M. With no

KCl, [H+]/K= 18.29. From [II4]/K = (l-o)/», it follows that o = 0. 0 52

and [II4] = QC = 6. 2 x 10" J; therefore, pH =5.21 and pK = 6. 47. The

value of pK is unexpected; however, it was reproducible in the vacuum

set up. The value of pK = 7. 2 as suggested in Fig. 19, and which agrees

with the literature is obtained in buffered solutions. Thus, for vacuum-

prepared solutions, we adopted pK = 6.47 for dilute KCl, say 0. 02 M.

The two-band method for 0. 020 M of Harshaw KCl yielded

[H+]/K = 22. 87. From pK= 6,47, pH = 5.11. Similarly, for0.022M

of RAP-KCl (URL, C-4), [H+]/K = 36. 0 5, or pll = 4. 9 1.'" If we let

[II ] represent the case of C - 0, mole fraction of Oil in the crystal,

it follows that.

7. 78 x 10'6 - [H+l - 0. 020 C.. 1 J o H

12. 26 x 10" ' = [H+l - 0. 022 Cn 1 J o R

This is also C for [OH (s)] since the solution is 1. 00 M in KCl.

The low values in pH will be discussed later.

57

- ■ ■' ■■-- - -■-■ ^----^-—^-^.■..- -...^-■.. ■-....--^- ■■ -■ .■■■.-■■ ..-^.: ^;..■...— .■■.. ■-■^-.-^-■■--■^■■.- ■■ : ^A.-i.. ..-1|i- ■ ■ . ,_.„,-.. ■ ■ ■ .■-■-■ -.^^■-:..l- ■ ^ ^^ h u ■.■ - ■

■

where C.. and C., are the mole fraction of OH (s) in the Harshaw and 11 K

RAP specimens, respectively. These two relations yield.

CH - 1. 1 CR = 2.24 x 10 -4

- 3 We believe that RAP growth can easily achieve a CR < 10 ; therefore.

C - 2. 2 x 10 -4

At the end of the contract we realized that there were still

problems remaining. How would one define [11 ] as a basis to

calculate C9 Is the low value of C due to RAP constituents dissolved

in the crystal? Obviously, the method needs to be optimized; e. g. ,

choice of the dye. The low pH values seen in this work surprised us as we expected

solutions of KC1 (or Nad) to be close to pH = 7. This was not the case.

Our pH meter, calibrated against standard buffered solutions, registered

a pH of 5. 8 to 5. 6 for 0. 3 M KCl (or NaCl), independent of the source

of reagent powders examined. The water employed to make the

solutions was deionized; when boiled and cooled under N2 it measured

to pH = .7. 0.

58

-- -'-^ : -'-■.—■'■»»■.-'-^■^--^—- ■.-.■^.... , min,- ■- ■■ -- ..-:-■■■-. ^ .,........... ■..-.,-. . ..„,. ., -—'--■■- --

?»W'9l!*??9B!PP"J"W*W?«'™TWlt,,»*»W!"WfflWPW^^ i" V ■■.■W.WÄ«w^mwrnw** fn. 1,1 iiw■ Jinw*|j*ti|ip*■.«.(^,»!PVFJ»-?PHJWIk^i.. 1..m 1 MI . .1,11. itj^

5. Other Preparative Approaches

a. Hot Pressing

Hot pressing produces a uniform size distribution of fine

grain material and, therefore, better uniformity in mechanical behavior.

Further gain in mechanical strength results from the use'of mixed

halides. The object of th<* study was to establish the feasibility of

applying the technology to mixed halides for high-power laser windows.

The anticipation of tcchnolAgical problems in hot-pressing

determined that the study of a one-component system should precede

that of the mixed halides; Potassium chloride was adopted as the

working material with (99. 9%) powder as the source. Pressing was

carried out with an 0. S-in.* diameter double-action tungsten-carbide die under Ar.

Purity control was not considered initially. The temperature-

pressure-time dependence of compaction under A*r, expressed as per-

cent density, %d, of the crystal was established at 2 50C. It was found

that %d does not vary for dwell times exceeding 15 min. Figure 10 shows

KC1 powder at 2 50C in the pressure range of 7. 5 x 10 to 3. 5 x 104 psi,

under a 15-min interval, compacts to a %d which ranges from 89. 0 to 99.6.

For the purpose of suppressing hydrolysis, a brief study of the

tradeoff between temperature and pressure was made. At 104 psi,

%d = 94. 0 at 2 5 C (refer to Fig. 20) while at the same pressure at

200OC, %d = 99.2. To obtain a %d = 99.2 at 2 50C, a pressure of 4

2.7x10 psi is necessary (refer to Fig. 20). With this tradeoff, press-

ing at ambient temperature is feasible, which is an important feature

for the suppression of hydrolysis.

Hydrolysis is favored by an increase in temperature of the

material. The equilibrium constant for the ionization of water (K ),

which process yields OH (and H ), increases 100% for every 130C rise

in temperature. Thus, the factor of three increases in pressure at 2

constant %d avoided a 10 increase in the concentration of OH". This

follows from, [OH-] = K UZ, in the ionization of water. The increase

59

^g. _-_______ ■-— ■■ - ~-->~--.^-^- ... ..--. . ,......-■.,,.■.■■ .„-: ..

100

98

96

^p 94

92

90

88

UNDER Ar

VACUUM

20

I03 psi

25 30

I4I9-SRI

35

Fig. 20. Dependence of Percent Density of KC1 Disc on Pressure at 250C (Pressinq Time - 15 m in).

60

_,.W>. !_■.,.,_- „

^HPl«"!"«^-'-.'''"^!» "I"»". -'"VJ-i^1 .i!.i,i ppi^UJWJPlWiHU^'n^itW.'iiW

in Kw from 2 5 to 200OC is, Z 175/1 3; therefore, for [OH-], the increase is ^ ' , or 10 . Also, the same increase in [H ], in the

presence of Cl , favors the formation of HC1, a gas. Hence, the

Le Chatelier effect operates to favor further hydrolysis.

When KC1 powder was pressed under Ar at 2 5 C, fogging of the

disc was observed. The phenomenon appeared different from that

commonly observed in the fogging of pressed KBr. In the latter case,

fogging is rapid, (within an hour or so) and is observed to be a surface-

layer process with minute cracks developing and propagating. In the

present case, it took about a day for the same degree of fogging (visual

rating) to develop. This observation led to the hypothesis that the

pieces were metastable solutions of a gas (Ar) in the solid (KC1) with

slow kinetics of exsolution.

The thin samples which were pressed under Ar at 2 50C and 4

5,0 x 10 psi for long periods of time, achieved %d = 100. The pieces

were clear and gave an excellent transmission under the infrared

microscope (1, 06 [im). One such disc, weighing 0. 38 g and with a

thickness of L = 0. 1 5 cm, gave ß = 0. 058 cm" and ß = 0. 057 cm at

tWO different beam locations (<2 mm apart). While these observations

make a good case for homogeneity, it should be pointed out that the

source powder employed is capable of achieving ß < 0. 0005 cm' in the

single crystal form grown by RAP. The above specimens eventually

fogged under storage. No control or exclusion of water vapor from the

air was exercised.

In line with the hypothesis of gas, fogging should not occur in

the disc obtained by pressing in vacuum. Hence, a 1-in. diameter

vacuum-pressing die was designed and fabricated. Subjecting the same

ii iterial to a 1 5 min evacuation, a curve of %d against pressure at 2 5 C

and a dwell time of 15 min was established. The resulting compaction

was significantly better than that carried out under Ar (refer to Fig. 20).

Using the same source powder, the discs from vacuum pressing

were found to be already fogged as they came out of the die. Obviously,

this was not the gas exsolution behavior observed earlier but a bulk

61

■^....■■.. ...........

■■ i***m*Kmimivpmß.

precipitation phenomenon. With a < 0. 1-mm Hg pressure maintained

during pressing, it seemed unlikely that precipitation was caused by

anions formed from atmospheric constituents but rather absorbed or

incorporated constituents, as in hydrolysis (OH"). It had been shown

previously that the extent of hydrolysis increased with the specific sur-

face of the material. xe ' ' J Also, such fogging may have been present

but not appreciated in the pressing under Ar because the pieces were a

factor of four thinner. On the other hand, Ar may be acting or reacting

in some manner which is yet to be understood. To check out the effect

of hydrolysis, a comparison of the unscrubbed and scrubbed powder in

vacuum pressing was pursued.

Thus far, hydrolysis has served as a heuristic model which has

worked out consistently as a guide to material processing approaches.

Consequently, the use of scrubbed powder (Section II. A. 2) was preferred

over a further study of pressing under Ar.

Vacuum pressing of the unscrubbed powder ranged from 25 to

150 C, 7. 85 x 10 to 7. 64 x 10 psi, and 5 to 90 min in dwell time.

These were 1-in. diameter discs. In all cases, the disc was already

■fogged as it came out of the die. This was not the case for the scrubbed

powder. Figure 21 -»hows a comparison of the transparency between the

unscrubbed (L = 0. 62 1 cm) and the scrubbed (L = 0. 626 cm), at 150oC 4 '

7. 64 x 10 psi and 60 min dwell time.

Three determinations of the ultimate strength (rupture) for each,

yielded 1780 ± 200 (3) psi fcer the unscrubbed and 2560 ±3 10 (3) psi for

Uie scrubbed. Whereas the microhardness (Knoop) of the unscrubbed -2 '

10 kg mm , was just about the value for the single crystal, the scrubbed • * -2

specimen gave 18 kg mm , which value approximated that of the

press-forged material.

Our measurement of microhardness on press-forged KC1 crystal grown by an earlier RAP procedure was 19 kg mm"2. (Procedure has been optimized since then. ) This was a specimen provided by R. W. Rice who obtained a yield strength of 5500 psi. See Semiannual Report No. 1 (30 June 1972), ARPA Order 2031, Naval Research Laboratory, Wash. , D. C.

62

.-.-.-. - - ----

" ' mmmm^mwmmi I I I- i. u»™ —. .M.I. ^^B I • i >vnKw>B^VW>wn-w>VPW^m«HOTI^^WP^^^^VVMr»>«^av-WVW^niM

15

ill

Scrubbed Uns crubbed

Fig. 21. Hot Pressed KC1 , 2.54 cm Diameter and 0.62 cm Thick, From Scrubbed and Unscrubbed MCB (99.9%) Granules

63

_......... .-^ ifciMaaiagmaiftHw^iMHi im i - - —- — - — -----

'»■■»•« ■'■ i ,...—.«,■•-« imtwrnwrnmîm «m m'vm,*mm.m< .iiin-in-w UMI^HP ^JÎ I I II .^^.^-..^ . , ■■m i.. >uw. ...... — . . ««.. «!■■■■■ .Li ( u-^^^^mnwi i« P ■ i\mmm»m, m,m m w. wfrnmî^m

For compai-ison, in the same Instron equipment, Polytran Rave a yield

point of 600 psi and an ultimate strength of .SOO psi.

The optical transmission and absorption coefficient at 10.6 |j.m

were: 46% and ß = 0, S5 cm" for tlie unscrubbed (weight : 6.22 g and

L = 0. 62 cm) and 84";. and ß 0. 077 cm' for the scrubbed (weipht^ 6. 24 g

and L = 0.63 cm). The one-order-of-magnitude improvement in ß attests to

the importance of the initial surface state of the starting powder. The

pressing parameters still need to be optimized. For instance, the

dependence of ß on thickness is appreciated, using the scrubbed powder.

For 1- ■ 0. 15 cm, both long and short dwell times, 18 hr at 2 5 C and

1, 5 hr at I10OC, save similar results, ß = 0.050 ± 0.001 cm' , which

is significantly lower than the case of L = 0.63 cm.

A mixed halide mechanical mixture of equimolar NaC 1 and KC1

granules, from MCB (99. 9";.), was scrubbed with CC1.. A sample was

hot pressed in vacuum using the 1-in. diameter die at 150 C, 5. 1 x 10

psi, and 60 min dwell time. The resulting disc was opaque.

Another mixed halide mechanical mixture, KC] and KBr, 1:2

molar ratio, was scrubbed with CBr. and CHBr,. A sample was hot 4 J „ o 4 .

pressed in vacuum using the 1-in. diameter die, at 170 C, 5. 1 x 10 psi,

and 60 min dwell time. The resulting disc was opaque. However,

KCl-ZKBr single crystals grown under an internally-funded program,

press-forged under Ar, produced repeatedly clear discs. The 1/2-in. 3 3

diameter die was used at 650 C, 2. 66 x 10 to 7. 64 x 10 psi, and 60 min

dwell time. Powder scrubbing and solid-solution crystal growth are far

from being an optimum RAP because one deals in this case with a mixed

halogen solution.

With the crystal, ß = 0. 002 cm' , which is a factor of two lower than the value reported by AFCRL. The crystal shows no yield point. However, there is more than a factor of four variation in the modulus of rupture. Our measurement yields 5400 psi while the AFCRL reported value is 1345 psi. It is not known at this point in time whether these differences are real and attributable to RAP. For the AFCRL results quoted, see Table 3 on page 6 of "High Power Infrared Laser Window Materials (LQ-10 Program)," Quarterly Progress Report No. 5 (1 Jan to 3 1 Mar 1972), Air Force Cambridge Research Laboratories, Bedford, Mass.

64

The results of our studies on materials purification,

topochemical exchange, and crystal growth from the congruent melt

lead us to believe that significant advances in the technology of hot

pressing will come from the development of a one-step RAP operation

from scrubbing to pressing. It is conceivable that such a development

can be easily realized with mixed halides based on a common halogen

(chlorine in particular). The case of mixed halides based on mixed

halogens will require more extensive background studies.

b. Crystal Growth From Solution

The object in this study was to obtain high quality single

crystals of high purity NaCl and KC1 with an OH'(s) mole fraction of -7

C < 10 . The exchange reaction of interest is,

CMc) + OIl"(aq) — OIl'(c) f CMaq), (9)

where (c) is the crystal and (aq) is the saturated solution. It follows

from the mass-action expression for eq. (9) that,

C = K lOHJMiL (10) [Cl (aq)]

If it is assumed that the ionization constant of water is pK = 14 even 1 w for the saturated solution, then

[OIMaq)] = lO"14 + pI1 . (11)

The saturated solution of KC1 or NaCl is approximately five molar,

cr

[Cl"(aq)] = lO0,7 (12)

65

It follows from eqs. (10) to (12) that,

in -14. 7 - (pK - pll) (13)

-7 From eq (13) it is easily demonstrated that, in principle, C < 10 is

quite feasible. Besides, Oil (s) can be controlled via the pH of the

solution; e. g. , adding 11C1. Depending on the value of K, there may-

be very little OH Is) even without intentional pll control.

Solution-grown specimens would provide standards for the

behavior of the melt-grown counterparts as these achieve low OH (s)

content, for instance, F-band coloration susceptibility and transition

temperature from extrinsic to intrinsic dc conductivity. These

expectations are justified to the extent that occlusion can be avoided

during growth. The main concern in obtaining high quality single crystals

from solution is control of the growth behavior. However, this feature

may not be independent of purity because of the effect of traces of ... 18

growth poisons.

A very low concentration of PbCl was employed as a growth

habit modifier for NaCl and KCl. Figure 22 indicates that there is an

optimum concentration of PbCl-, required in solution growth. Figure 23

shows the crystal quality obtained with NaCl and Fig. 24 for KCl.

The transmission at 10. 6 |J m of specimen (a) of Fig. 23 with

thickness L = 0.41 cm was 93%, specimen (b) ranged from 53 to 74%,

and specimen (c) from 0 to 15%. Specimen (a) was repolishod and its

transmission from 2. 5 |jm to 20 |j m (Beckman IR 121 was compared

with the melt-grown materials, as shown in Fig. 2 5. It is quite obvious

that the IR transmission of the solution-grown material is inferior to

that of the melt grown. Of the melt-grown materials, the RAP crystal

is the better transmitting material. A comparison between the two

extremes, the solution-grown versus the RAP melt-grown NaCl, was

also made in terms of the optical absorption coefficient at 10. 6 pm, ß,

using an adiabatic calorimeter apparatus (refer to Appendix II). The

solution-grown NaCl gave ß = 0.013 cm" while the RAP melt grown

yielded, ß = 0.0018 cm' . Most likely, the higher ß in solution grown

66

■ II ■! « ■■■- - '- i i ii mir ^ '- ä- *^^ ■ -.- -^

iuuww«M,.i%:*ww*l""wwi'Jl^ '■'^.^■"^■'■"'■^"^w.^w'^^^w'PPipi!^^ .niiifivppRiiipRppiiiiPMiiiui niag.: ..•vmnim WffPWPWiPB^ÎBiWWPWpW^WPIWWWlil^

(a) Spherical growth of KC1 from solution 1% "wt PbCl2

(b) Crystals of NaCl showing high opacity in the center, which was the initial growth with no PbC^. The clear growth was effected with 0.251 wt PbCl2

Fig. 22. Influence of PbCl? on Growth From Solution (Magnification l.bx).

67

.... -....■.-... I ■MlM^^lM^^^H^^^^^ÎIHllMMMÎ «III I

"■ -i'-Jl »1 WI*HP^^WpP!*TWi"'w-^- ■J." >JH^ Mil||HWi-U ■■! II ■! V*iai|ilUIWUÎ*VI I'll MWIII Illl inH^MBW-.I^MrHWi ■ IWi«âmR«mHM.mMI ■«"■■■ I» WMM» J «IM ■ U.I ■««■■» MP I ■■■I.Î

la;

From left to right, the crystils were grown as follows: (a) Convection cell, x^PbClp) = ?.3 x lO-3 , and pH 1. (b) Convection cell, xlPbClp) = 1.0 x lO"4 and pH 2. (c) Beaker at 70oC with x(Pb(N03) ) = 1.8 x lO"4.

Fig. 23. Solution Grown NaCl Crystals (scale unit: inch) .

58

mimm'mm?'*tt<mmwJn'.^wrTmm**mi»*mmy. "■vum'vmmw'm**'*'''''' '"•l,' • - -''■>"''mm." ". «•>-™pi«»pm.ini."i"»iUJIMI.HIJII!.,I, m^mmmm

Clear outer layer was grown in a 4

solution with /(PbClg) = 1.8x10 and pH 2.

Fig. 24. Solution Grown KC1 Crystal (scale uni t: inch) .

69

'— MM •MMH-HaaMMMOMB ^^aMHMM.

■MOTM m^^m^mmi

I o 01 10

I

E

z Ul _l UJ

I

o s-

a> s: -o

to

o •r-

■t-J

o

o t/> (/)

•i—

E

c

T3

LT)

CT

NOlSSIINSNVdl lN30ä3d

70

_'.'_.. _ _^.. _ .. ^,r ..i.....:^,^,^,. -* I ^11 ^...^„■.^^..^-^^....„^^w .■^..-..,.„ n j-jrfJlMriiilVan -■■->-'->^-r-^ '-' ■.-■■'.^~-.^^^-

^..^.„...«'■■^l.. ,...^

NaCl is due to occlusion and other growth defects that plague solution

growth. These drawbacks were not rectified by scrubbing with CCl.(g)

at 670OC for 20 hours, which yielded ß = 0. 010 cm" .

In the case of the solution-grown KC1 specimen (refer to Fig. 24),

the internal region of the crystal was cloudy. The polished specimen

yielded an 85% transmission at 10. 6 |j.m. A thin section (L = 0. 27 cm)

was sliced off, poHshed, and gave a 93% transmission. The larger

section (L - 1.2 5 cm) gave an 87% transmission. In view of the results

obtained with NaCl, and the slightly better quality of solution-grown

NaCl over KC1, further comparison of solution-grown KC 1 with melt-

grown KC1 was not warranted.

71

"^■•^■'"'•''"^^iW'WJWff^

B. WINDOW COATINGS

!• Theoretical Considerations

a' Design of Antireflection Coatings

Antireflection coatings can range from a simple single layer

having virtually zero reflectance at one wavelength to a multilayer system

of twenty or more layers having virtually zero reflectance over a wave-

length range of several octaves. The type used in any particular appli-

cation will depend on a variety of factors including the substrate mate-

rial, the wavelength region, the required performance, and the cost.

For design of coatings and for tutorial purposes, most reference

works assume that the substrates and materials used for coating are

completely transparent (i. e. , k 0, where k absorption index or

extinction coefficient of the material) in the spectral region in which the

reduction of surface reflectance is to be obtained. Cox and Hass 19

and H.A. Macleod,20 using this assumption, have presented

detailed designs and discussions of anti reflection coatings for optical

surfaces using a single layer, double layer, and multilayer coatings.

Therefore, this section will not include any detailed discussion of the

derivation of these designs since they are treated in an excellent man-

ner in the above works, although material and design information for

10. 6 Km is not included in those references. However, design data

incorporated in this section does consider the factor of absorption in

film materials, because at the present technological stage of coating

application methods, the high power 10. 6 Mm laser flux to which the

coatings will be subjected makes the inclusion of absorption considera-

tions in the designs mandatory.

Practical coatings are generally restricted to one- or two-layer

designs with an occasional use of a three-layer coatihg. This is par-

ticularly true at 10.6 (xm where the large physical thicknesses required

in the layers cause problems in the realization of a given design.

Single layer coatings are limited in usefulness since one obtains a zero

72

. ^ p ll..-i..-...-i..',^.i.-. . ^■■...■.;.J^.J.J...,i|irjni^ ■■.^,.;,..■■

■ m m -~ « N ■■ I Uli I • !• . i ,1 i i.awuwiipiiMi ii •■•wiH>^nMi^mi^npp~>iiiii.iiiiiiiijii ■ iman.iHiiiia lav

of reflectance only when the refractive index of the coating material is

equal to the square root of the substrate index. When this condition

cannot be met, two or more layers must be used to obtain a true zero

of reflectance using available materials.

When two layers are used in the coating design, the added free-

dom permits one to choose many different combinations of refractive

indices and layer thicknesses to produce a zero of reflectance at a pre-

scribed wavelength. The basis for choice of a particular design to pur-

sue experimentally then becomes criteria such as the minimization of

absorption and the minimization of differential strains which might lead

to poor adherence. In the work reported here, minimization of absorp-

tion has been the prime criterion in making the choice of designs.

An important aid to designing two-layer anti reflection coatings

is a diagram such as that shown in Fig. 26 for ZnSe (see Ref. 19).

Every point in the shaded regions of the diagram represents a pair of

refractive index values fur the coating materials which will produce a

zero of reflectance on the given substrate. Points outside the shaded

regions, but close to them, will yield low reflectance but not zero.

The corresponding thicknesses are then calculated from a pair of

design equations. The final step is to calculate the 10.6 fj.m absorption

expected for each coating design based on either measured absorption

indices for the individual film materials or best guesses where no

measurements are available.

The remainder of this section contains some illustrative coating

design information for KC1, CdTe, and ZnSe. This information is

strictly theoretical in nature, and there are many factors that contribute

to the actual performance of a coating that are not included in the theory

leading to the design thicknesses and absorption values. Therefore, the

optimum experimental thicknesses need not be the same as the theoret-

ical ones (although they are usually expected to be close) and the absorp-

tions need not be as low as the theoretical predictions. Some of these

problems are discussed further in the section on experimental results.

73

..-. i'tiiirttiriihiVlfcfiiiii^'iirlii rm1 -^^■^/■j/^r-^^...

wtM«p!lWiillippjliUJ4ll( mii*mmmm*imm*mr*i*mmm***imm*m!immm*i&mimrewmmmim 111 l,MH-Ul lniljJWl4ll|Uijuu«N)jmil.nni,ll,lnirll

1941-6

-- 3 -

UJ

<

UJ

o X UJ Q 2

2 3 INDEX OF OUTER LAYER, n-

Fig. 26. Double-Layer Antiref1ection Coating Design Diagram for ZnSe.

74

""^^ "■'■■•■ --■■ ■ ■ ■■ ---. • - ■ '■■■ ■' ■--■;■■■■ -■■ ■■ ^^~-.--— ..-.-,, ■■ ..i........

■"""-"■T* ii.mi ,i,.«iiiuu»rai«HiiwsB»5w^wppwiM»ijiiiiiiiui jiiiiiiijiii wmmmm "i" ^^mmm

b- Antireflection Coatings for KC1

In the case of a low index halide, like KC1, there are no

available coating materials with refractive index equal to or less than

the substrate index. This means that single layer coatings and double

layer coatings from the rectangular region or the lower triangular

region of a diagram for KC1 similar to that of Fig. 26 are not feasible.

The possible two layer coatings come from the region where the refrac-

tive index n1( of the first deposited layer is greater than the substrate

index ns. and the index of the second deposited layer is greater than n and less than n, ATP. s

1 N s Table VII contains layer optical thicknesses as a fraction of the

10. 6 ,.m design wavelength and predicted absorptions per coated sur-

face for a variety of designs in which the second (outer) layer is chosen

from among the low index fluorides ThF^ IfeF, and Sr^'which have

low absorption at 10. 6 ,.m and low water solubility. The As S /ThF

combination has beên extensively studied experimentally and^sults tor

it will be reported. A detailed picture of the dependence of total absorp-

tion per surface in this coating on the absorption in the individual layers

is given in Fig. 27. The point labeled 1971 in the figure was shown at

the 1971.Laser Window Conference asjhe expected best absorption in

such a coating based on the best absorption indices which had been mea-

sured in As2S3 and ThF4 to that time. The point labeled 1972 is based

on improvements in the two materials during the intervening year.

If the refractive index of the first coating layer is high enough,

the potential range of the second layer index can be increased to include

ZnS as the second (outer) layer. This occurs for the combinations

ZnTe/ZnS and CdTe/ZnS as shown in Table VIII. These coatings incor-

porate no low index fluorides and. therefore, should be less water sen-

sitive than those in Table VII. Table VIII also illustrates another

important point in the design of two-layör antireflection coatings. This

is that, in general, there are two independent solutions for each pair of

refractive index values (optical thickness values for one may be sub-

tracted from one-half wavelength to find those for the other). The two

75

^■■■•;,-■■ ■•■• ■— ^ -■'■^■-■■-'- "-^- ^" -- ■-*■"-■■ ■ .-. ....,..., iULLWî ..,.-.■„/.^Û.lT..^...„,..-... L......

p-iF-'ii.,itJ*u.| .P-^-.^-PllillHII 1UHJU.*MJP1II.UI igillillllHB^ll^nniHp^MimP iilli! ijiiAJUli ß iiiM'f^J" I-" i .Ulili Jiuii. "HIiHllfJW.IllJ^iJllll.HllJllJIJI 11,1 jiimji.,iii.iWL,iMJ|ii

I263-3RI

0.10

0.08 a

Q:

cr

^o 0.06

| 0.04 en CD <

0.02

10

As2S3 ABSORPTION INDEX

)f9.3xlO~5

|0-H

ThF4 ABSORPTION INDEX

5x10"

2x10 -5

10 -3

Fig. 27. Absorption per Surface in As2S3/ThF4 Antireflec- tion Coatings on KC1 Versus Material Properties

76

■ -■ - - '"'- —■'— ■'■■■ ■■'-■■- ■■. ■ .„..,.■-....—„.-., .................*„—..... . ^^. ^^ ji-ni i n i-YtTtiMi

TABLE VII

Theoretical Designs for Antireflection Coatings on KCl

\Layer 2

Layer 1 N^

ThF4 BaF2 SrF2

n « 1.5, k = 1 i

x lO"4 n ' 1.395. k = 5 x lO-4 n = 1 36, k = 2 x 10"3

Ll L2 A*(%) Ll L2 A*(%) Ll L2 A*U) ZnS n = 2.16 k = 2 x lO'4 0.412 0.166 0.066 0.437 0.168 0. 142 0.445 0.172 0.423 AS2S3 n = 2.377 k = 6.5 x IQ"5

In2S3

0.432 0.157 0.032 0.450 0. 163 0.105 0.456 0. 167 0.378

n = 2.4 74 k = 6.5 x 10"5

CdTe

0.438 0.154 0.172 0.454 0. 161 0.251 0.459 0. 166 0.526

n = 2.67 k = 1 x lO-4

ZnTe

0.447 0.150 0.040 0.460 0.158 0.111 0.465 0.163 0.379

n = 3.0 k = 1 x lO-4 0.457 0.146 0.039 0.467 0.156 0.109 0.471 0.161 0.376

Absorption per c oated surface.

T728

TABLE VIII

Antireflection Coatings Without Low Index Fluorides for KCl

ZnTe/ZnS

" = 3-0> A n = 2.16 k = 1 x IG"4, k = 2 x lO"4

CdTe/ZnS n = 2.67, n = 2.16 k = 1 * 10 4, k = 2 x lO"4

Layer 1

High Index Thickness , Wavelengths

0. 116

0.384

0.185

0.315

Layer 2

ZnS Thickness, Wavelengths

0.343

0.157

0.297

0.203

Absorption, %

(Per Surface)

0.054

0.049

0.053

0.051

■T52Ö-1

77

—-J ii umiiiiiiiii

1 "■ '■■" ' " »W^^RW^—»PW mm > '■ '■

solutions each will produce a zero of reflectance at the design wavelength

but differ at other wavelengths, and they need not have the same absorp-

tion. In the case of the designs of Table VIII the predicted absorptions

for the two solutions of each set are nearly equal so a choice between

them could be made on other grounds, such as minimization of differen-

tial strains. For the coatings of Table VII the choice of solutions has

been made to minimixe predicted absorption where BaF? or SrF is the

outer layer. In the case of ThF4 outer layers the predicted absorption

for the two solutions was almost the same - the choice was made to

limit the 'IhF, thicknes'-

The ZnTe/ZnS coating with optical thicknesses 0. 116 and 0. 343

wavelengths was chosen for experimental study, and results will be

reported in the Section on coating results achieved.

Figure 28 shows more extensive theoretical predictions of the

10. 6 |jm absorption per surface for this coating as a function of the

absorption in the individual layers. The point on the figure represents

the current best-guess values for the two absorption indices in the

absence of firm experimental numbers (taken as equal to that achieved

for CdTe films for SnTe, ten times bulk value for ZnS).

A novel coating design, which is a modification of a basic double

layer anti reflection coating of possible use where the outermost layer is

sensitive to moisture in the environment, consists of adding a thin third

layer of a film material having low optical absorption at 10. 6 |a.m and

good resistance to moisture. Such a design, for example, ]s to add a

thin third layer of CdTe to a basic two-layer CdTe/BaF design. The

properties of such a three-layer coating in theory are as follows.

CdTe/BaF on KC1 where the thickness of the CdTe and BaF 2

films are 0.46 wavelengths and 0. 153 wavelengths at 10.6 |j.m, respec-

tively, will theoretically have reflection = 0 and absorption in percent

per surface of 0. 1 11% as shown in Table VII. If the two-layer design

above is modified by adding a thin layer of CdTe, the CdTe/BaF /CdTe

design with film thicknesses of 0.46/0. 10/0.017 wavelengths at 10.6 |xm

will have a reflection of 0. 074% and an absorption per surface = 0. 077%.

The CdTe outer layer physical thickness = 675 A.

78

- - - - • - ---..-■■■-J. ■ ,| ■ —iMia- - - ■ - * .. . .-.

Mi.*iwi«»,)ij*"ii-*i,-,"y ■rvnmnimmmmimm -- "- in i iiiHii i^i» -P^«—i^ mmnrnw^miM^m ■-•■« «WfH«|pipf7«PH|PMUl^i» —"'^^^^^

0.10

0.08 - UJ ü

12

10 tc 0.06 üj a

O 0.04 »- a. tc o vy CD < 0.02

1941-5 I I I I I I I I I I I I I I I I

10

ZnS ABSORPTION INDEX

I I I I I 'i I i i

8 |54 2 4 6 4 6

ZnTe ABSORPTION INDEX

Fig. 28. Theoretical Absorption per Surface in ZnTe/ZnS Antiref1ection Coatings on KC1 Versus Material Properties.

79

.-..■.—-.,...--L ■:-.,...■,■^.,.^.^.,.. ■_ ,. .■■■ , ^..1..^_^.>.J::J..J-,-...-... ■ ir - l1|||Mi¥t|Mrii;, MM «r ill ' llli IIMMMB Mil if' I ■ r....^-. ,^-....... ., „', „^.....^L.^.^..^,.....-

i.|i.iiM.«'iW»-WJMVJ'^^'"■"«■■" >>•-*.*wrr*~n '-,V^ W-? ^ff^W^WW ■ 11'- L..JI iLMJJWIl.ai.LU! UIBUii'nilMiW*»''

c. Antireflection Coatings for CdTe and ZnSe

In the case of these semiconductor window materials

their high refractive indices relative to that of the low index halides

provide new options in coating design. Single layers are theoretically

feasihle if zero reflectance is not required. Also, double layer coat-

ings are possible with refractive index values in all three regions of

the design diagram. In particular, one may now choose double layer

coatings from the rectangular region of the diagram where the index of

the first layer is less than the square root of the substrate index and

the index of the second layer is greater than the square root of the sub-

strate index. These coatings have important pro] crties which are illus-

trated below.

The design diagram for ZnSe is shown in Fig. 26; the diagram

for CdTe would be similar. The three numbered points on the diagram

correspond to the identifying numbers in Table IX, which contains

design data for these three coatings as well as for a ThF single layer.

At each point the two-layer solution which minimizes absorption and

coating thickness has been chosen for inclusion in the table.

A comparison of the solutions labeled 1 and Z illustrates the

advantages of the solution in the rectangular region of the design dia-

gram (No. 2). They both utilize the same materials to produce a zero

of reflectance at 10.6 |j.m — but in opposite order. For solution 2, which

has the order BaF^/ZnS, the required coating thicknesses are much

less than for solution 1 (which is important at long wavelengths), and

the predicted absorption is less than half that for solution 1. Another

important advantage of solution 2 is that the water-sensitive low index

fluoride is not exposed to the environment; the outer semiconductor

layer protects it.

Solution 3 in Fig. 26 and Table IX (Ge/ZnS) is included because

the Ge/Znh corrbination represents an important pair of materials which

is widely used in commercial infrared optical coatings. The use of

germanium may be prohibited at high power levels because of the danger

of thermal runaway, but a Ge/ZnS antireflection coating for ZnSe may

still be usable for low power applications.

83

" i -■ ■ ■ ...^ny^.^o^nj j -"'-■■'- ■ ■' - ■'■ —' "- '•■' " --■■■—'■'flii'untr ittMirn ! ■' ■ •■■^■fa

■ "''-"»^^T-T^ajnrrw^rww^^wwi-Pwr^^^w^ WmtW-mx&WW^mrr WPI,artJ^m^H.«|H|.l|«B|:iU^pW»:JU^,WMn»!.WJII^IWi^

No. On Desi gn Diagram

TABLE IX

Theoretical Designs for Antireflection Coatings on ZnSe

Coating Materials

Layer 1

ThF(

ZnS

BaF,

Ge

Layer 2

ZnSe

2.42

BaF,

ZnS

ZnS

Layer Optical Thicknesses , Uavelenaths

Layer 1

0.250

0.222

0.138

0.116

Layer 2

0.241

0.059

0.303

10.6 um Optical Constants

ThF,

1 .5

1 x 10"

BaF, ZnS

1 .395

5 x 10 -4 2.16

2 x 10'

Absorption Per Coated Surface,

%

0.023 (R = 0.13%)

0.142

0.066

0 043

Ge

4.0

2.5 x 10'

T729

81

■ ■ - ■ — —■ - - — — -' — ■-■ ■ ■—-— - -

Table X contains design information for some typical coatings

on CdTe. Again one sees the striking improvement in performance and

reduction in layer thicknesses that can be obtained with coating designs

from the rectangular region of the design diagram (compare ZnS/BaF

with BaF2/ZnS). The ThF4 single layer and the /.nS/ThF and

naF-,/ZnS double layers have been studied experimentally; results will

be reported below.

2. Coating Results Achieved

A general introduction and background to optical coating tech-

nology that included a number of theoretical designs and predicted per-

formance levels for coatings on KG] and CdTe windows were presented

in previous sections. It^is appropriate at this point to summarize what

has been accomplished to date experimentally in efforts to prepare opti-

cal coatings for these materials. The primary goal was to try to mini-

mize the 10.6 fxm absorption in the coatings. In order to assist in

achieving this goal, measurements were made of the 10. 6 (jan absorp-

tion in the substrates, and of the coated substrates so that the coating

absorption per surface, as well as the absorption index of the film,

could bo calculated. The film absorption index values were important

to obtain so that experimental results cuald be compared with theoret-

ical designs.

a. Passivation and Antireflection Coatings on KC1

Substrates for our coating work have included KC1 single

crystal substrates grown by Optovac raid Harshaw, as well as I1RL

grown single crystals. In addition, coating studies were performed on

Harshaw polytran substrates received from AFWL. The halide coating

work placed emphasis on the development of single layer passivation

co tings and two-layer antireflection coatings.

The passivation coating results are summarized in Table XI.

To dais, the single layer As^S, passivation coatings have shown the

lowest coating absorption per surface, if they are prepared on KG]

82

Ig», MMMHMn iHiilftWMIMIMMi - ■ •■'---•-"

TABLE X

Theoretical Designs for Antireflection Coatings on CdTe

Coati ng 'lateri als Layer Optical Thicknesses, Wavelenaths

Absorption Per Coated Surface,

% Layer 1 Layer 2 Layer 1 Layer 2

ThF4

ZnS

ZnS

BaF2

ThF4

BaF2

ZnS

0.250

0.133

0.185

0.119

0.207

0.212

0.076

0.024 (R - 0.73%)

0.034

0.127

0.058

Note: 10.6 ,iiii optical constants: same as Table III for coatinn materials, n = 2.67 for CdTe

T730

TABLE XI

Properties of Passivating Coatings on KC1 Windows

Materi al I ndex Thi ckness , Wavelengths 1 , ,0

A, % Per Surface

ThF4

ZnS

CdTe

A52S3

1 .5

2.16

2.67

2.377

V4

V2

V2

94 t 1

94 i 1

96 1 1

94 ± 1

0.52

0.13

0.13

0.02

T731

83

iiMHPum lima

substrates that have good surface quality and are not ha/.ed or fogged as

a result of reaction with moisture in the air. For example, where the

substrate absorption was high (because of poor surface quality

0. 0123 cm ), \/2 As^S- was 0. ZZ% per surface. With substrate absorp- 1 2 3

tion of 0. 00125 cm a \/2 As?S coating absorption of 0. 02% per sur-

face resulted. It should be pointed out that in using ThF. that its thick-

ness is not limited to X./2 for passivation but can be deposited much

thinner (thereby reducing absorption). This is because o^ the close

match in the refractive index of ThF, with KC1 which are 1.5 and 1.455, 4 ' respectively.

Antireflection coatings prepared in our work, which to date have

emphasized As?S_/ThF. two-layer coatings, have continued to yield

percent absorption per surface greater than the desired goal of 0. 1% or

less (see Table XII). The As?S- was deposited by e-gun evaporation

and the ThF, by resistance heating of tungsten filaments. The thick-

ness of both layers was optically moni'-ored by a He-Ne laser. Fig-

ure 29 shows the spectral transmittance measured on a Beckman IR-12

spectrophotometer of one such sample, in comparison with a theoretical

plot which utilizes the same parameters of index and thickness for each

layer as the experimental data. However, the increase in absorption

associated with ThF. above 10. 6 |jLm is not included in the theoretical

plot which produces high transmission values as seen in Figure 29.

The dip observed between 1600 to 1700 cm in the experimental data

is also attributed to ThF.. It is believed to be caused by OH in the

ThF, and can be reduced by substrate heating during deposition. Calor-

imetric absorption measurements point out the necessity for this reduc-

tion for it has been found that increases in absorption at 10. 6 \iTn can be

correlated to the depth of this dip.

Because of the problems which seem to beset ThF , some non-

fluoride antireflection coatings were tried. The most successful has

been a two-layer ZnTe/ZnS coating which to date has yielded absorption

values of 0. 05% per surface (less than the desired goal of 0. 1%). Both

layers were deposited by resistance heating and optically monitored at

a wavelength of 1 fj.m. Preliminary investigation of this coating design

84

- ^k - -

^"■^~«">p^""~"" "'' ■ ■ ■ ■ ' ""—^"■•■^■»

1941- 4

O 60

< tr

40

20

THEORETICAL

EXPERIMENTAL

ThF4 O.I6X0

As2S3 0.43X0

\

KCI SUBSTRATE

J 1 1 1 1 I I I l I 1700 1500 1300 1100

WAVENUMBER, cm-1

900 700 500

Fig. 29. Spectral Response of AsoS^/ThF- Anti refl ection Coated KCI . ^ J 4

85

- . .- . .

led to a value nf 3. Q for the refractive index of ZnTe, compared with

the value ofZ.7 1 reported by S. O. Smith And J. S. Seeley . The

coating design for the ZnTe/ZnS data shown in Table Xll utilized the

original index value of 2. 71, resulting in a peak transmission from the

Beekmen ipectrophototneter of only 95%.

b. Experimental Coating Results for CdTe

The three different antireflection coatings that have been

studied on CdTe windows are ThF. single layers, and ZnS/ThF and 4

RaF-)//.nS double layers. The dominant experimental fact is that for

all three of these coatings the theoretical design data does not match the

experimental results. For the ThF. single layer of 1/4 wavelength

thickness, reflectances below 4.5% could not be achieved (theoretical

prediction 0. 73";,) whereas, for the double layers minimum reflectanre

is realized for layer thicknesses relatively far from the theoretical

ones. The reasons for these discrepancies are not known, but they are

believed to be a result of surface effects on the CdTe. Similar effects

should be watched for.on ZnSe, which might be expected to have surface

properties like those of CdTe.

Table XIII shows results to date for the three coatings studied

on CdTe. The coatings were deposited in a conventional diffusion

pumped vacuum system with liquid nitrogen trapping onto CdTe surfaces

heated to 150 C. The ZnS source was a vitreous carbon crucible with a

tungsten pancake filament, the BaF, source was a heavy Mo boat. ThF c. ' ' 4

was deposited, using both types of source, with no apparent difference

in performance.

"The ThF4 single layers were monitored with a 10.6 fim CO,,

laser so the optical thickness is known ^.o be 1/4 wavelength at 10.6 fj.m

in all cases. Physical thicknesses were measured interferometically

to confirm the assumed value of 1. 50 for the ThF. refractive index at

10.6 )j.m. The minimum reflectance achieved for these 1/4 wavelength

layers was consistently in the range from 4. 5 to 5. 0% in conflict with

the theoretical prediction of 0. 73%.

86

i iii^ai i ■ -^—^^■^»^^»»■■" ■ ■■' ," ' ■" "»« ■•—■■■ i i

TABLE Xll

Coating Results on KC1

Mul ti 1 ayer Coati ng

Beck man Spectrophotometer

Absorption Pet Surface Percent

Peak Trans. Percent Predi cted Goal Achi eved

As2S3/ThF4

ZnTe/ZnS

10.6 urn

10.5 MM

99 t 1

95 i 1

0.03

0.05

S 0.10

? 0. 10

0.19

0.05

Windo1-/ - HRL's RAP Grown KC1 Single Crystal

T732

TABLE XI11

Experimental Antireflection Coating Results on CdTe Windows

Coating Desi gn

ThF,

ZnS/ThF4

BaF2/ZnS

Theoreti cal Opti cal Thi cknesses {Wavelengths)

0.250

0.133/0.207

0.119/0.076

Theorot i cal Reflpctance

0.73

Experimpnta1 Thporotical Reflectance 10.6 .MI ,ib-

( ) Sorption per surface ( )

1.5-5

0.1S-0.2!

0.15-0.30

0.021

0.034

o.o 5r

E xperi menta1 10.6 nil absorption per surface ( )

not measured

0.25-0.50

0.06-0.25

T733

87

- ■

^^*^mm^m*m ii ■"I" MI .. ^—■ "" '■ !"■ ■ -' . ■um "

The ZnS/ThF4 double layers were also monitored with a CO?

laser. In this case it was found that using the theoretical design thick-

nesses the reflectance at the minimum was between 1 and 1.5%. The

thicknesses were then adjusted empirically to obtain minimum values in

the range of 0. 15 to 0. 25%. The final prescription arrived at for the

deposition was as follows: deposit ZnS until the original CdTe reflec-

tance of 20. 7% is reduced to 10. 3%, deposit ThF until a minimum is

reached. The theoretical stopping point for the first layer is 13.9%.

The BaF /ZnS double layers were monitored with a IIe-\'e laser

at 6328 A. The theoretical design thicknesses require deposition of

2. I wavelengths of BaF^ and 1.4 wavelengths of ZnS at 6328 A taking

account of the known dispersion in the two materials. The use of these

thicknesses produced a reflectance of 1. 1% on the CdTe windows. When

the thicknesses were empirically adjusted to 1.75 wavelengths at Ci32S A

for both layers, reflectances in the range of 0. 1 5 to 0. 3% were consis-

tently obtained. Translated to 10.6 um, these results imply that mini-

mum reflectance is occurring at optical thicknesses of 0. 100 wave-

lengths and 0.0^0 wavelengths for BaF and ZnS, respectively; whereas,

the theoretical values are 0. 11 Q and 0.076 wavelengths. Figure 30

shows the spectral response measured from 2. 5 to 22 ^m for a 2-in.

diameter CdTe window anti reflection coated with the empirically adjusted

BaF^/ZnS design along with a theoretical curve calculated from the data

shown on the figure. The reflectance at 1 0. 6 fim was also measured with

a CO^ laser; it was found to be quite uniform with an average of 0. 73%

(two coated surfaces) across the face of the window. The 10.6 [im

absorption measured calorimetrically at the center of the window was

0. 25% per coated surface.

3. Pulsed Laser Damage

Some preliminary data have been obtained for the behavior of

coated and uncoated alkali halide windows when subjected to high power

C02 laser pulses. The test results are shown in Table XIV. All tests

88

■■Ml ll I

— "^«MW^WIW

I88I-»

100 -

i* ao

^) 60

to

< ■ 40

20

1 1 1 - —f 1 1 1 1 1

THEORY

• • • EXPERIMENT

t

\ /.

<<" "^^^

fc

\ / / • •

• • THEORETICAL CUPJVE CALCULATED WITH: •

~ \ / *

1 / * \ / *

n OPTICAL THICKNESS

BoF2

ZnS

1.41

2.16

O.IOXQ

0 09X0 Xo-'O.S^

• •

" —

1 1 1 1 1 1 II 1, !

10 12 14 IÖ

VVAVELEfiGTH, /in 13 20 22

Fig. 30. Spectral Response of BaF2/ZnS Antireflect! on Coating on CdTe.

89

uaa_aia_BM>A —

—^^•^^^^mm^m pw^iW»P^^^^^WPw^^^WPWP»ww«wnwi ^^mt^

TABLE XIV

Pulseu Laser Damage Results

M indow Substrate

Coating Absorption Measured Damage^ - Threshold (J/CB ]

NaCl (Harshaw)

None * 40*50

KCl «87-21 (Polytran) pol> crystal line

None

•/2 As2S3

0.0016 cm"1

0.02 /Surface

14

6-9

KCl »69 (Optovac) single crystal

AR type

As2S3/ThF4 0.8 /Surface 5

KCl «B-24-6 (HRL) B r i d g m a n

None 0.00076-0.00080

en"1

46-57

grown

not measured

• 30

TTJ?

90

■

were performed on the URL fast-flow test bed using a multimode stable

resonator with a pulse length of 2. 2 ^sec. The power output of the laser

was adjusted below the threshold for clean air breakdown using a 6.4 cm

G« lens. The sample was located in a converging portion of the beam

With the highest intensity upon the exit face. Since the spot size is

determined by the lens to sample distance, corrections are included

for the effect of the index of refraction of the sample upon the exit spot

size. The test results are based upon a single pulse damaee threshold,

performed upon vinstressed windows.

4- Surface Finishing

To date, no single material has been developed to the point

where it exhibits both the low optical absorption and the high tensile

yield stress necessary to fulfill the requirements for a window in a

high-power laser system. However, since'the required optical perfor-

mance has been achieved in single crystal KC1 grown by reactive atmo-

sphere process (RAP) at IIRL the discussion on surface finishing will

be limited to this material with a u-u remarks on the polishing procedure for CdTe.

a. KC1 Finish

Grinding and finishing relatively soft, water soluble

materials, such as the alkali halide to a finish that does not degrade

the intrinsic optical absorption of the window material, requires spe-

cial techniques. The two most significant problems encountered in

polishing halide materials are caused by their relative softness and

their solubility in water. Therefore, they are difficult to clean, and

the finished surfaces are unstable in normally encountered environments.

Attaining surfaces of high grade optical finish is very difficult on such

soft materials because of the strong tendency to develop sleeks and

"orange peel" during the polishing process.

The technique utilized in this work to achieve a high quality

finish on KC1 consisted of a two-step process. Samples were initially

91

.

ground on a cast iron lap with SiC emery paper (No. 600) to remove

gross surface roughness. They were then finished dry on cloth using

the abrasive Linde A (Al 03) and a small amount of ethyl alcohol.

With this technique, two types of surface finish were given to

KC1 samples. They are designated as optical finish and window finish.

Optical finish samples are worked until flat and parallel within one

fringe of visible light. Window finish samples are flat within a few

fringes, with no attempt made to achieve parallelism.

Prior to coating, slices from various locations in an URL single

crystal boule of KC1 were polished and subjected to calorimetric absorp-

tion measurements. These results are shown in Table XV. Although a

correlation with sample length or position in a boule is not apparent, «

strong dependence on the type of surface finish exists. The optical

finish samples showed a strong tendency to fog during the few minutes

that they were exposed to the 50% relative humidity laboratory environ-

ment before placement in the desiccated calorimeter, even though placed

in desiccators immediately after polishing was completed.

The most striking example of the effect of a fogged surface on

the measured absorption is slice Xo. 3, which was measured with both

surface finishes. It was first found to have an absorption coefficient of

0. 00048 cm with a window finish. After return to the optical shop for

application of an optical finish, the absorption coefficient was found to

be 0. 00185 cm . The higher absorptions measured in the fogged opti-

cal finish samples may be a result of surface absorption caused by the

fogging, but they are also partially due to the high scattering of these

surfaces. Light scattered directly to the thermocouple in the calorim-

eter causes a temperature rise in the thermocouple which is then attrib-

uted to absorption in the sample. This phenomenon can be defected by

examination of raw data. It is manifested as a component of the temper-

ature signal, which does not have the normal time lag due to heat con-

duction when the laser is turned on or off.

To determine the nature of, and reasons for, fogged surfaces,

two slices with window finish were coated with approximately 100 A of

evaporated gold to prevent surface charging and then examined for

92

--■ - - ■ ■ —■^___-^__-,—^^—_„__ .__ _...

wmmmim^mK^mm

Slice No.

3

7

9

12

TABLE XV

Absorption Measurements on KC1 Samples

Length, cm

0.97

0.S7

0.89

1.00

0.83

Finish

Wi ndow

Optical

Optical

W i ndow

Optical

W indow

Absorption Coefficient,

cm'1

0.00048

0.00185

0.0014

0.00050

0.00145

0.00115

Comments

Not fogged.

Fogged while exposed for placement in calori meter.

Slightly fogged while exposed for placement in calorimeter.

Not fogged.

Slightly fogged whi le exposed for placement in calorimeter.

Near Br-ule heel possible-Impurity contamination

' ~ T691-1

93

m ii ■hnni^ mini - ■ ■-

structural differences in the scanr.ing electron microscope (SEM).

Figure 31(b) shows a fogged surface It is evident that the fogging of

the surface is caused by rec rystallization of the surface region of the

crystal; the upward growing crystallites scatter light and thus produce

the fogged appearance. Figure 31a shows an unfogged surface. This

may be compared with the underlying surface of Figure 31b which

appears at approximately the same magnification.

From other examinations, it appears that the surfaces of optical

finish and window finish samples are very similar in structure. The

optical finish is, however, smoother and more uniform, with a charac-

teristic size of the features a few tenths of a micrometer, similar to

that of the Linde A abrasive used to prepare it. The window finished

surface has some regions with the same characteristic size as the opti-

cal finish but also has some features greater than 1 [im in size. The

number and depth of scratches is also greater on the window finish

surface.

Aside from indicating the need for further work on the reduction

of surface roughness, the scanning electron microscope examination

has not yielded any clue as to why the optical finish fogs so rapidly.

Efforts to improve the quality of the surface finish for single

crystal KC1 windows have been emphasized, because as we have pre-

viously reported, nearly all of the KC1 samples polished at URL had

fogged surfaces shortly after polishing. Recently, this problem was O

brought under control by heating the KCI windows to =50 vith an 1R

lamp immediately after polishing. Though it is believed this process

removes water that was present on the surface as a result of the alcohol

polish, it is not clear that this is the only process which leads to the

increase in surface stability. The window can then be placed in a vac-

uum deposition chamber for coating or potted in carboline plastic for

future use. Because of the hygroscopic nature of these halides, we

find it is advisable to remove the carboline from the potted windows in

a heated environment to avoid any reabsorption of water.

Investigation of surface roughness has also been emphasized.

Scanning electron microscope studies on polished KCI windows show

94

-- i — ■

(a) Unfogged window finish

(b) Fogged window finish

Fig. 31. Scanning Electron Micrographs of Polished KC1 Surfaces.

95

=-

surface features of a few tenths of a micron, which is characteristic of

the particle size of the polishing abrasive, Linde A, Unfortunately,

dielectric coatings not only reproduce surface features, but amplify

variation in the surface contour, making it even more desirablo to

reduce roughness to a minimum. In an attempt to reduce the roughness,

Linde ß, whose particle size is a few hundredths of a micron, was used

as the polishing abrasive. This resulted in a smoother surface finish

(see Figure 32). With these improvements, very consistent results

have been obtained with absorption coefficients of 0. 0015 cm' for opti-

cal finished and 0. 0005 cm for window finished samples. As indicated,

the optical finish (flat within one fringe in the visible with parallel sur-

faces) results in a much higher absorption coefficient than the window

finish (flat within a few fringes with no attempt at suifacc parallelism).

b. CdTe Finish

Because no major effort was placed on the study of sur-

face roughness, cleaning and absorption as a function of surface finish,

only the polishing procedure to achieve a high quality finish on CdTe

windows will be discussed.

The windows are initially ground with Al O, (20 |j.m and 9. 5 |j.m

particle size) and then polished on a pitch lap using water as the vehicle.

Linde A as the starting abrasive, and completing the procedure with

Linde B. Typical calorimetric absorption values ranging from

Q - 0.002-0.005 cm were obtained for these windows.

The results achieved in window coating technology to date do not

reflect a limitation based only on the coating state of the art itself. The

performance of the coating is greatly limited by the substrate surface

finish quality prior to application of the optical coating. The KC1 sur-

face finish quality is limited by the present status of optical firishing

technology and the pronounced tendency for KC1 to haze and fog upon

exposure to the environment prior to coating.

However, as the progress reported in this report indicates,

further improvements in surface finish technology should continue with

96

the use of polishing agents with small grain si^e (10' (im or less)

along with a surface cleaning procedure prior to the application of opti-

cal coatings. These .mproved finishing techniques will result in less

surface damage, smoother surfaces, and minimum optical absorption

at 10.6 ^.m for high power laser windows.

lt4l-7 1141-8

(a) Polished with Linde A (b) Polishea with Linde B

Fig. 32. Scanning Electron Micrographs of Window Finished KC1 Surfaces.

97

S E C IM O N III

CONCLUSIONS AND RECOMMENDATIONS

A. WIND' W MATERIALS

As stated in Section 1, the object in the study of window materials

was to upgrade the technology of metal halides for their use for high-

power lasers. Although the metal fluoride technology was given some

attention, the emphasis in the study was on alkali metal chlorides (in

particular, FC1 and NaCl) for high-power laser windows at lOfjm.

Consequently, the conclusion and recommendations which follow deal

with the study of alkali netal halide technology. Although the study

centered on KCI, the te-rhnology is just as applicable to NaCl.

The main thesis of tiie materials study was that the characteriza-

tion measurements by earlier workers for high-power laser window

applications; e.g., optical absorption at 10.6 ^m ( ß), mechanical

strength (yield, Y, and ultimate, S), surface stability, etc., were not

intrinsic to the halide materials but reflected the status of materials

technology at the time. The measured quantities were parameters and

not properties of the materials as to be shown by their dependence on

the history of the material, (i. e. , processing). We conclude from the

study that the proposition has been demonstrated.

At the start of the study, window grade commercial KCI showed

P : 0. 007 cm' . The development of a one-step RAP technology now

produces reproducible KCI crystals up to !0-cm diameter with

P < 0.0005 cm' . The inequality sign is to draw attention to the limitation

arising from the technology of optical finishing rather than ot materials

puri fication.

Also, at the beginning of tlve study, the mechanical parameters

characterizing the measurements on the current KCI crystals were:

Y ^00 psi and S 640 psi (Rcf. 1). The one-step RAP technology,

applied to purification and crystal growth as well as annealing, have led

93

-o an increase in Y by more than 100"',, and in S by as much as SnO"'.,.

A similar improvement is indicated in the application of a parallel

technology to metal fluorides (see Appendix VI).

The (Y, S) of the single crystal are considerably higher than those

of the large-grain polycrystalline aggregate (Polytran) but low, especially

in Y, compared to the very fine-grain aggregate obtained by hot pressing

or the press-forged form. Although water etching shows the feasibility

of achieving in single-crystal KC1, a more than one order-of-magnitude

increase over the old value of S, it also shows that a value of Y < 10 psi

appears to be an intrinsic limitation of the undoped material. However,

the single-crystal approach based on congruently melting solid solutions

of mixed halides, demonstrates the feasibility of achieving Y=6 x 10 psi

or values of the yield point comparable to the press-forged metal halide.

Solution-growth procedures did not produce crystals that served

to provide the standards for the melt-grown crystal. The study showed

that the procedures given in the literature for the solution growth of

high-crystal quality alkali metal halides fell far short of meeting the

performance requirements for high-power laser windows. Continuing

the study of solution growth for laser window applications is not

recommended.

One can say that adeqi ate control of anion purity, especially the

exclusion of OH (s), has been established in the study. RAP recipes

for materials purification, crystal growth and annealing have been

developed. Although the study treated the measurement of [OIMsl] in

the last quarter, a satisfactory procedure remains to be developed and

tested for reliability. The importance of such a procedure lies in its

use as a control of, and a monitor for, materials processing of metal

halides.

The study also demonstrated the need for a greater depth in the

investigation of the physico-chemical aspects of surface fabrication and

finishing. The achievements of the one-step RAP served to emphasize

the inadequacy of the current technology of surface fabrication. We

regard the study of the latter as one of the frontiers in the upgrading of

halide materials technology to meet the rigid requirements for high-

power laser window applications.

99

^

The upgradinp of halide materials technology must, of necessity,

proceed by a series of approximations. It has been the good fortune of

tins study to have established early a recipe for the reproducible - 3 1

achievement of p < 10 cm (KC I single crystal). Subsequently,

upgrading of (Y, S) became possible without exacting an exorbitant

degradation in p. Of these two mechanical parameters, the primary one

is Y. The upgrading in S is required in the sense that S >Y to the extent

that the material is far from being brittle.

The one-component single-crystal approach upgraded S to a far

greater extent than Y, maintaining p<IO cm'1. The plastic region

widened but Y fell far short of the minimum bound, 6 x 103 psi. Our

recommendation for further study, based on the single-crystal approach,

is the investigation of single-crystal solid-solution mixed halides. I his

appears to be the most promising approach which can meet the homo-

geneity requirement in scale-up. An early feasibility study of the growth

of large crystals should be undertaken.

For the approach based on the use of polycrystalli;ie agcregates

as laser windows, press fabrication appears to be the most practical.

Press forging should start with crystals grown by the one-step RAP.

This investment in the quality of the starting material is a necessity

and will avoid complicating the process further. It is inherent in this

process that the problems of homogeneity, (e. g. , grain size distribution)

will gain prominence with scale-up. For this reason, it is also recom-

mended that the study of hot pressing, a process capable of a high degree

of uniformity, be continued. This asset, which stems from the use of

a powder with very small grain size, is at the moment also a drawback.

The high specific surface of the material and the problems of interfaces

are no doubt the causes of the degradation in p. The study demonstrated

very significant improvements, indicating the necessity of developing

a one-step RAP for the process.

100

B. WINDOW COATINGS

The major effort on the window coatinp task was directed toward

the development of technology for producing passivation/anti reflection

coatinqs on laser windows that would have absorption at ID. 6 |jm of

0. 1% (or less l per coated surface. This p;oa! was attained for coatings

on KC1 and on Cd Te substrates. The KC1 coating consisted of a two-

layer coating of ZnTe and ZnS, and resulted in an absorption of 0. 05%

per surface. The CdTe anti reflection coating consists of two film layers

of BaF-, and ZnS, and resulted in an absorption of 0. 06% per surface.

Another coating design which was investigated for KCI, was a two-layer

combination of As^S and Th F which gave a best value of absorption

of 0. 19* per surface, (somewhat higher than the program goal require-

menti. The absorption in the ThF film was the limiting factor in 4

this latter ease.

The performance of the coatings on KC1, in terms of their

optical absorption and mechanical and surface passivation qualities,

are strongly dependent on the techniques used for KC I surface finishing

and the various process steps used for fabrication of the KC1 window

from the starting KC1 powder to the final window blank. Optimum

performance of the finished and coated window depends on stringent

and careful quality control of all procedures and processing steps used.

The level of performance we have achieved to date is a result of our

focusing attention on the following procedu^'s:

(1) Reactive atmosphere processing of the KC1 crystals in a one-step RAP from powder to final crystal boule

(2) KC1 boules are saw-cut to approximate window blank thickness with a wire diamond saw using Convoil fluid as a lubricant.

(3) The rough surface grinding operation is performed on a cast iron lap with SiC emery paper (No. 600) to remove gross surface -oughness

101

(4) The final surfaces are finished dry on cloth using Linde A (AKO-j) abrasive and a small amount of ethyl alcohol, until the required level of flatness and parallelism is achieved. (These window finish samples are flat within a few fringes of visible light, with no attempt to achieve optimum parallelism.)

(5) Directly after polishing, the window blanks are heated up to 50oC and placed in a vacuum deposition chamber for application of the optical coatings or potted in carbofilm plastic for future use

(6) The optical coating thicknesses during coating deposition are carefully controlled using optical monitoring. Spectrophotometric transmission curves are taken of the coated windows to insure that coating thickness are accurate and to enable corrections to be made to the process if transmission of the coatings is not peaked at the JO. 6 ^m design wavelength.

Continued effort is required to develop surface finishing and

coating technology for halide and semiconductor laser windows for use

in high power infrared laser systems. The ultimate achievement that

is required is to provide a technology for window surfaces and coatings

with high-power laser damage thresholds as close as possible to the

intrinsic threshold of the bulk window materials. The results achieved

in window coating technology to date are still limited by the present

coating state of the art and by the substrate finish quality prior to

application of the optical coating. Further improvements in surface

finishing technology should be continued with the use of polishing agents

with small grain size (10 fim or less) along with investigation of

surface cleaning procedures to be used prior to the application of the

optical coatings. The improvements in the surface finishing techniques

should result in less surface damage, smoother surfaces and minimum

optical absorption at 10. 6 \ini. The important aspect of future finishing

work should be to develop a technology which will also permit the

achievement of the window flatness and parallelism needed for opera-

tional laser systems. In our work to date we have not been able to

meet the latter requirements without suffering some increase in

absorption in the KC1 window over the intrinsic material property.

102

APPENDIX I

REACTIVE ATMOSPHERE PROCESSING OF ALKALI HA LIDES

Ii one conslderi what mechaniimi may be reiponsible for

excess tO-|Am optical absorption in alk.ili haiick'S, the hydroxyl ion

lOil i coupled to a lattice metal ion looms as a major possibility.

Since OH has its source in water, a ubiquitous material, it is likely

to be present in relatively large amounts unless it is stringently con-

trolled. Achievement of such control was, therefore, a major initial

effort in early phases of the high power laser window programs at

Hughes Research Laboratories.

Contamination with OH occurs through hydrolysis in the pres-

ence of water vapor .it the elevated temperatures necessary for crystal

growth. This process is summarized in the simple reaction equation,

written here for chlorides.

Cllct • H90(g) * OH' (c) • HC1 (g| 14)

which tells us that in the presence of water in the gas phase (g) a

chloride ion in the condensed phase (c) is replaced by a hydroxyl ion

With the production of HC1. From the equilibrium relation for this

reaction, we find that the ratio C, of hydroxyl ion to chloride ion in the

sol id is given by:

c i°ILi£ll K- [CMc] 1 P

ILO

IIC1 (15)

where K. is the equilibrium constant and the p's arc partial pressures.

103

Since the objective is to achieve a low value of C, one must

remove the IÔ and provide HC1. The use of vacuum or an inert gas

is not wholly effective in achieving a small C because PJ^Q is never

zero, wirle P^ is essentially /.ero in those circumstances.

The first alternative that suggests itself is to use an HC1

atmosphere, a procedure which has been effective in the analogous

problem with fluorides. However, the halogen acids are corrosive

and in a flowing system, corrosion of the ducts by HC1 provides a

means for contamination. This disadvantage is overcome in a closed

system. But, if such a system were to remain compact and held to

ordinary pressures, the amount of HC1 available for exchange is very 1 imited.

The use of a noncorrosive and a condensed system, which can

act as an exchange agent, overcomes these drawbacks. Since H O is

a component that cannot be easily excluded, it is best to adopt a reac-

tion that converts lÔlg) to HC1 to favor a low value of C in eq. (,'>). Examples are

SOCI2 I H20 -- SO, • ZHCt1

COC^ i ll20 ^ C02 • 2HCI (16)

CCI4 • 2H20 - co2 • 4HCI

Each of these compounds exhibits a corresponding exchange reaction with OH

molecule of CCI4 consumed is in the crystal. With CCI^ the exchange cycle per

20H"(c) • ccîg) ^ 2cr(c) • C02(g) • 2HC£(g) (17)

104

t^t^tâmtmtmam

There is an advantage to the use of CC1 in that thermal

dissociation occurs at < 300OC, yielding two reactive species,

CCiA - CCI3 I C^ dg)

There are five possible reaction paths for the products of thU dissocia

tion. One obvious path is the reverse of eq. (IK). Two paths are accounted for by

2CCI3-C2CI6 (H)

and

2C^ - Cl2 (20)

Both products in eq. (19) and eq. (20) serve as indicators of the pro-

gress of the reaction. The ^C^ deposits as fine needles on the

colder parts of the apparatus, and Cl2 imparts a yellow color to the source (liquid CC1 ).

The remaining two paths are the scrubbing reactions of CC1

and Cl on the crystal's surface layer. Quite likely, the deepest pene-

tration is effected by Cl. The atomic radius of Cl is ~0. 94 A while

Cl is 1. 81 A. In RAP applied to the crystal, its diffusion into the

bulk may also be greatly enhanced by a resonance-type electron trans-

fer, Cl + Cl"(s) - Cl" + Cl(s), etc. In such a case one does not have

to rely on OH" diffusing out all the way to the surface for the exchange reaction.

Among the halogens, chlorine has the largest electron affinity

(unit: eV): F, 3. 58; Cl, 3.81; Br, 3. 54; I, 3. 29; and OH, 1.73. This

feature provides the drive for the electron transfer process,

C£(c) + OH"(c) - C£"(c) + OH(c) (2i)

105

The lormation of molecular chlorine from the atomic form, eq. (20)

preserves the potential for exchange:

Cl2(g) i OH"(c) a* C\~{c) + HCl(g) I 1/2 ©-(g) (22)

It should be noted that the use of Cl in RAP includes the

advantage derived from the use of HC1, as shown in eq. (14). The

reason for this is seen in eq. (22) which shows that HC1 is a product

of the action of Cl^ on OH". By the same token aside from the corro-

sion factor, the use of CC14 is an optimum because of the benefits

derived fiom the generation of nascent (atomic) chlorine which in turn

can produce molecular Cl .

Equations (14) and (22) are not independent. They are related

through a well known reaction (the Deacon process):

Ce2(g) t H20(g) ^ 2HCl(gJ < l/202(g) (23)

It follows that a form like eq. (15) can be derived with

K, - K25.,K22 (24)

where the subscripts of the equilibrium constants at the right hand side identify the respective reactions.

The second degradation which shows further advantage in the

use of CC14, is the formation of phosgene by a scavenging action on

HO from outgassing of the apparatus,

CCl4(g) * H20(g) m COCl2(g) i 2HCl(g) (25)

Phosgene, in urn, hydrolyzes readily to form more HC1:

COCl2(g) f H20(g) - C02(g) 4 2HCI(g) (26)

106

^MMMMM

Th« parameter PHJO^HCI1 '^ ^•c^,ive ^•ckor ^ •^ '151' is

sensitively dependent on either eq. (25) or eq. (26), because Z moles

of HC1 are formed at the expense of i mole of I^T^*

The reactive atmosphere is effective at all stages of crystal

preparation, purification, crystal growth, and anneal. The use of

reactive atmosphere processing (RAP) adds another component to the

system. Naturally, the equil ibrium behavior (e. g- , solid-melt) deviates

from the the rmodynamic behavior reported in the literature. For

instance, la) the melting points are observed to be higher, (bl solid-

solid transitions in one-component systems disappear, (c) incongruent

behavior reported for certain binary compounds (two-component halides)

is lifted, id) composition is shifted for congruent behavior of solid

solutions of mixed halides, etc. These changes in physical behavior

are accompanied by changes in the measured physical properties" of

RAP crystals. For example, optical absorption at the l-pm and 10-fj.m

regions, thermal conductivity, electrical conductivity, mechanical

hardness, yield strength, ultimate strength, etc., effects important

for high-power laser windows.

The process parameters of KC1 and NaCl allow the merger of

purification and crystal growth into a one-step RAP (refer to Sec-

tion III. A). The study demonstrated that a fairly cheap starting

material (less than a dollar a pound), 99. 9"'o pure or mole fraction of

metal ion impurities x =: 10 , can be employed. Most likely, the

process can utilize a lower grade starting material, but this is not

an advantage. The longer processing time incurred by the use of lower

purity starting materials provides the limiting factor. The case just

considered is dynamic RAP (flowing system).

For purposes of scale-up, compact configurations are desired.

Hence, it is of interest to determine the order-of-magnitude upper

bound in x which renders feasible a closed-system RAP crystal growth.

With no sacrifice in generality, let MO be an oxide impurity (say, in a

KC1 melt) and consider the following scavenging action,

MO(c) f CC24(g) ■• MC^2(c) + COC£2(g) (27)

107

Again, it is instructive to write the equilibrium constant, K^, as

follows,

I PcOC22 fMO(c)1 _ [kci2(c)j KZ7 PCCI

(28)

Obviously, to achieve a very small value of the ratio on the left-hand

side of eq. (28), a large value of PCci4 should be applied. In a large

scale operation, this choice is not too practical in a closed system. In

a sealed-off quart:-, apparatus, PCC14 ■ 150-mm Hg at room tempera-

ture, or a vapor-phase concentration of C = 8 X 10 mole cm .

The other alternative to displacing reaction (27) forward is:

Moles CC1. » Moles MO, i. e. ,

cV » nx , (291

where V is the volume of the vapor phase c is as defined in the

preceding paragraph, n is the moles of KC1, and x is the mole fraction

of the impurity MO.

It follow.? from equation (29) that.

x < < c m) ■ (30)

-2 where v is the volume of the melt. For KC1, n/v s 2. S X 10 " mole

cm"3. For crmpact operation in a closed system. V/v ~ 3, which

approximates the volume ratio of the powder charge to the melt. There-

fore irom eq. (30), the tradeoff to a closed-system RAT crystal growth

operation is that: x<< 10' , i.e., the starting material should be at

least four-nines pure.

108

APPEN DIX II

OPTICAL ABSORPTION OF POTASSIUM CHLORIDE AT TEN MICRONS

The calorimctric absorption coefficient at 10. 6 ^m, 0 , hns been

measured in NaCl, KC1, and KRr (Ha rshaw opt^al crystals,. '

A linear fit of fractional power absorbed by the sample, P /P , against

■ample thickness, L, extrapolated to L 0, Indicated a la'rge'surface

absorption 2., in each case. ' The behavior was attributed to the

hygroscopic character of the crystals.^ " We report here an assess-

ment in KC1 of the effect of surface hydrolysis and hygroscopic

behaviors on o, and a study of the dependence of P /p on L. The

effects on o of bulk purification, annealing, reworking'lhe surface and

the polishing vehicles used, are also presented.

Single crystals of KC1, 1- to 2-cm diameter, were pulled from

the melt using a vitreous carbon crucible (Beckwith Corp. ). The start-

ing powders were Matheson, Coleman, and Bell (MCB) 99. 9"o pure and

Johnson Matthey (JM) 99. 999%. In one of the runs the starting material

was a .one refined ingot from the MCB source." The steady-state

growth rate was 5 1 cm hr"1. Only the steady state sections of the

crystals were employed for the measurements described below. Sec-

tions of various thicknesses were obtained by cleaving at (100). The

surfaces were polished using a suspension of Linde A in ethanol.

The optical absorption of the material was measured using the

optical calorimeter technique.24 The technique is shown schematically

in Fig. 33. The sample is irradiated with power P. from a CO laser,

and the resultinR temperature rise with time AT/At'is measured2 The'

optical power absorbed by the sample P is given by

I!

1 - R)(l -

(1 - Re" 4 I P (31

109

- -

w*i^*w^*mm ■■*' .■^■ — ■^■■< —■<!■■■■■■ m-.». i ^ mt^

1858-3

Pi MONITOR

Jl L

RECORDER

4> SHUTTER ATTENUATOR L-J PT'Pi

SAMPLE M0NIT0R

HOLDER

Fig. 33. Laser Calorimeter Schematic

no

tmi~. - n -- -- -- — — ■

SS which is th« expression for the power absorbed by a sample of thickne

L, absorption a and surface reflectance R where R (n-l/n-l)2.

The expression is derived by assuming multiple surface reflections and

summing for multiple passes through the sample.

If this expression is then expanded in a Taylor's series, the result is

P. -H^t)^ ■■•■■] (32)

Fo r

/j_J_R\ oL u - R| 2 << 1 33)

the absorbed power can be approximated by

P ~ oL P. a i (34)

This treatment assumes no interference effects; i. e. , the sample is non-resonant.

For the case of a sample with parallel faces and normally

incident beam; that is, when the reflected beams interfere, the expression used for the absorbed power P is

a

-m 35)

where PT is the power transmitted through the sample. This equation

is derived in Ref. 24. From the calorimeter data, the absorbed power is

P B mC —— a At (36)

111

.

''•■, ■ ■ ■'

where m and C are the sample mass and specific heat. The absorption

coefficient is obtained from the calorimeter working equation, where

o mC LP

AT At

(37)

P. or i

ni_l 2n T

(38)

For the study of the effect of surface hydrolysis on surface

absorption at 10, 6 fjm, KC1 random cuts (Harshaw Chemical Co. ,

Cleveland, Ohio) were used. The equilibrium constant, K, of

hydrolysis.

Cl (s: H20(g) - OH (s) HCKg)

is given as follows.

[OH"(s)]

[CMs)] K

DHC1 (40)

The symbols (s) refer to the surface layer where the anion concentra-

tions are represented in brackets and (g) to the vapor phase where the

concentrations of molecular species are given as partial pressures.

To favor the displacement of reaction (39) from right to left, the crystals

were subjected to the halide gas (flowing system with He as the carrier

gas) at 500OC for 16 hours and cooled to room temperature under the

same atmosphere. The crystal specimens for the measurements were

generated by cleaving at (100) and polished initially with Linde A in

ethanol. For the first specimen, L - 0. 68 cm, Pa/P0 before and after

the treatment was 21. H x 10"4 and 22. 4 x 10' , respectively. Similarly,

for the second specimen, L = 1. 01 cm, the respective values were

112

^ ^ _ -■- - - -

24. 2 x lü-4 and ZI, 2 x 10"4. It is concluded that under ambient condi-

tions surface hydrolysis provides a negligible contribution to surface

absorption at 10. 6 pm. For the study of crystal surface corrosion by water vapor,

define PJJ Q as the ambient vapor pressure at a given temperature,

and P and P°T ^ as the vapor pressures of the saturated solution II2O H2O

and the solvent, respectively, at the same temperature. The three

regions of interest are:

(A) Dewpoint or Precipitation Region, where PH Q 5 PH Q

(B) Deliquescence Region, where P^ 0< PJJ Q "^ PH O 2 2 t

(C) Efflorescence Region, where PH Q ~ PH O

The consequence of a short residence time in (A) will be taken up in a

separate publication. In (B), a deliquescent behavior of KC1 or NaCl

can be conditioned by impurities, although the pure crystals are not

even hygroscopic at ordinary pressures and temperatures. In (C),

the KC1 surface should remain stable indefinitely if its surface free

energy equals that of the material employed to saturate the solution.

In general, this equality is not realized and the inequality can go either

way. Consequently, the study of hygroscopic behavior in (C) was chosen

and is described in the following.

A large bell jar housed the optical calorimeter (crystal and

reference). Then temperatures of the vapor phase and the humidostat

were monitored and heat lamps were positioned outside the bell jar.

Although a dynamic measurement of o by eq. (37) was not possible in

this setup, resetting the calorimeter in the optical bench and starting

a measurement could be realized within 5 min. A saturated aqueous

113

solution with excess KC1 powder was used as the humidostat. Vapor

pressure data show that saturated KC1 solution maintains an «0%

humidity from 20 to 90OC (75% for saturated NaCl solution); l, e. .

essentially, the same heat of vaporization as in pure water.

The deviation in the values ol Q shown in Table XVI is ±10%;

the steps are listed in chronological order. During heatup. incipient

clouding was observed in -10 min, at which time the gas-phase temper-

ature was 40 C. This condition never progressed further, even with a

3-hour exposure at 650C and PJ^Q - HO mm. Such surface condition

persisted for the duration of the steps that followed. If it affected a,

the effect is small enough to be beyond the gi^ven precision and, as the

data suggest, possibly reversible. These results rule out hygroscopic

behavior as the cause for surface absorption in the KC1 specimens under

study.

The results of absorption measurements at 250C for three

samples of KC1 given in the sequence (L in cm, 104P /P , N) are

listed below; N stands for the cumulative number of cleavages at (100)

for the preparation of the specimen.

SamPle 1 (2-05. W.«. 2). (MO. 15.4, 3), (0.95, 14.3, 3),

(0.56, H. 96. 4). (0.51. 9.69. 4). and (0. 27, 4.05. 5).

SamPle 2 (2-l0. 10-5, 2). (1.15. 8.05. 3). (0.89. 8.01. 3).

(0.66. 8.58, 4l. and (0.49. 7.35, 4).

SamPle 3 (2-52. 23.9. 2). (1.90, 22.8, 3). (1.26, 11.5. 3).

(1.22. 14.6. 3). (0.61. 9.15. 4). and (0. 57. 8.55. 4).

The value of a, calculated from eq. (37). is derived from

th(

as follows;

Pa/Po' the total Power absorbed with the contributions broken down

■L i o

20- i N.A < i (41)

114

...

TABLE XVI

Absorption Coefficient at 10.6 iim Versus Humidity and Temperature Exposures

Sample: HRL Czochralski Grown KC1 Crystal; Cleaved, Scrubbed, and Polished in CH OH-Linde A

Cumulative Time,

hr

Vapor Phase Temperature, 0c

P H20' mm

a ,

-1 cm

0 25 10 0.0044

+ 2, 5 25 19 0. 0044

+ 60. 0 25 19 0. 0046

+ 3. 0 65 110** 0. 0046

+ 3. 0 25 10 0. 0048

+ 15. 0 25 10 0. 0042

* Assumed to be the equilibrium temperature of the crystal.

Temperature of the humidostat (saturated KC1 solution) was 580C.

After a 1 hour cooling to match crystal and reference temperature to ambient {250C).

T566

115

whore i runs through all the spccLmens of each sample. The sub-

script 0 refers to the thickest specimen with N l\ Y is the bulk

absorption optimized for the given s.miple; i. e. , I® 0 and A 0.

Surface absorption; treated as a lonstant of the sample, is given by 2».

From the preceding studies, surface hydrolysis and hygroscopic

behavior, lv is assumed to be negligible.

After cleaving at (100), subsurface damage, in the form of

slips at (110) propagated into the bulk, are seen in the specimen

observed between crossed polaroids. Using a given cleaving procedure,

it is assumed that the contribution to bulk absorption per cleaving may

be characterized in eq. (41) by a constant A for each sample. A thick

cleaving tool was used for Sample 1, and a thin single edge ra/.or blade

for Samples Z and ^.

Other contributions to the right-hand side of eq. (411 can come

from surface reworking and the chemicals used in polishing. In the

present case, these contributions were held down to values beyond the

,jrecision of the measurements in order that one may examine eq. (41)

for the case A 0 (Model I) and the case of 2«r - 0 (Model II).

Were A 0 and 20" - 0, (P /P ), versus L. would yield a zero a o i i

intercept. The bulk absorption coefficient is given by the slope,

v /L . Each sample has essentially a constant geometrical cross 7 o o section Therefore, v /L will contain contributions to the absorption

' o o arising from sources distributed homogeneously in the bulk; e.g.,

dissolved impurities, scattering centers, etc.

In Model I (A 0), the least-square treatment of the data

(P /P ). and L., yields 2a- and y /L . In Model II, the least-square a o i i o o

treatment of the data, (P /P )., L., and N., yields A and y /L clOll I ()l)

Usinc A one obtains (P /P ).', the power absorbed corrected for the ■ a o i

cumulative contribution of damage by cleaving:

(P /P ). a o i ;p /P ). a o i N.A

i (42)

116

The U-n'it-square treatment of a linear dependence of (P^PJ^ on L.

servos to test the assumption that 2» a 0. The slope corresponds

to o ., which for consistency should be close to the value of fQ/l*0-

The plots of (P /P ). versus L. (Model I), and (P./PJ.' versus L i a o i i • u

(Model III are shown in Figs. }4 and ^5 for Samples 1 and 2 which

typify two extremes in behavior by Model I. A comparison of the two

models for the three samples is given in Table XVII.

Sample 1 shows the lowest damage parameer A; consequently,

I good lineir fit results with both models (see Fig. J4), Still, there

is an improvement from Model I to II of a one o rder-of-magnitude

reduction in Z« . Compared to Sample 1, Samples 2 and 3 show an

increase of about two orders of magnitude in A. Consequently, a large

Zff results in Model I (see Fig. 35). A value of Z« of ±0. 3 x 10" may

be taken as zero within the accuracy of the measurements. The entries

are given one digit beyond the significant figure merely for comparisons

between the two models. Note in Sample 2 that a large 2<T is accompa-

nied by a fictitiously low fj^ ^ UoAel I It is seen that jro/Lo

increases from Model I to II; the latter approaching from the low side

of tlv experimental value iPj^JJ^' For Li Lo' cleaving damage

contributes the least because N{) 2.

The improvement in the value of the bulk absorption coefficient,

y /L , with purification of the starting material in crystal growth is

scen'in a comparison of Samples 1 and 2 (Table XVII). Good agreement

is seen between yjl^ and oc in Model II (Table XVII). The correla-

tion coefficient to a linear fit improves from Model I to II (Table XVII).

In summary, the results support the two assumptions of Model II, 2<T

is negligible and A can be treated as a constant of the sample.

The results of annealing (50 hou- treatment at 700OC) Sample 3

specimens also support Model II: KCJ fabricated specimen (cleaved

and polished) with L ' 1-90 cm gave from eqs. (33) and (37)

o : 0.0012 cm"1 before anneal and a 0.00063 cm" after anneal;

another piece with L 1.22 cm gave a 0.0012 cm"1 before and

a -- 0. 00053 cm"1 after anneal. The values after anneal are almost a

117

__a^M -- -

1858-2

• MODEL I

A MODEL IZ

L,cm

Fig. 34. Optical Absorption of KC1 (Sample 1) at 10.6 pm Versus Thickness: •Model I. AModel II. '

118

15 1858-1

O

a ^

• MODEL I

A MODEL H

1

s s

^-•' s K/*

-

s

'' 1 1 1 1 • 1

5 —

L.cm

Fig. 35. Optical Absorption of KC1 (Sample 2) at 10.6 urn Versus Thickness: • Model I. A Model II.

119

■ ■■ - ■ --- , *- ^

■ ilrmriilwlHHH^m

-■ 1

MP^^^RW *■!

TABLE XVII

Coinpanson of Modell I and II in KC1

Meas urement

2:-10H

(yo/L0)-104, cm-1

'C-I0 ,cm

A-104 cm"1 cleave"1

Correlation Coefficient

Model 1

0.09

15.6

0.993

Sample 2'

2o • 1 0

(yo/Lo)-104, cm'1

.c-104, cm-1

A-10 , cm cleave"


6.69

1.71

0.903

Sample 3'

2"-10

-1 iyo/LQ)-]0*, ci

-c-104, cm"1

A-104, cm"1 cleave"1


3.53

8.58

0.957

Model II

0.01

15.7

1

0.01

0.993

0.19

3.73

3.51

1 .44

0.985

■0.23

9.16

9.38

0.85

0.969

Sample 1 grown directly from MCB (99.9^) KC1 powder. Sample 2 was grown from the ingot obtained by a three-pass zone refin^ ing of MCB (99.9C.). Sample 3 was grown from JM (99.999%) KC1 powder.

T675

120

—

n factor of two lower than the value of o for Sample 3 (Table XVII) and

c arc in line with the results of Sample 2.

The effect on o of surface reworking, as shown in Table XVIII,

indicates that the measure value may still have been affected by damage

introduced by the polishing operation. Such damage :s easily observed

using polarized light on an annealed specimen, before and after polish-

ing. Thus, it is very likely that with Improved techniques in polishing, -4 -1

a value ot o tor KC1 < 3. 5 x 10 cm can be obtained. c A genuine surface absorption effect can arise from the type of

liquid vehicle employed in the polishing operation. This effect was

demonstrated in a cleaved KC1 as follows:

Step 1 Faces were smoothed with 3/0 emery paper and

poiished with Linde A in ethanol: o 0. 0016

±0.0002 cm"1 and L 0.92 cm.

Step 2 Faces again were roughened with 3/0 emery paper

and polished with Linde A in glycerine-H O (1:1)

saturated with KC1 at pH < 1 (using HC1): o = 0. 0060

±0.0002 cm'1 and L - 0.90 cm.

Step 3 Faces roughened with 3/0 emery paper and polished

with Linde A in ethanol: o

and L = 0. 90 cm.

0. 0016 ±0. 0001 cm'

The source of the surface aosorption, which accounts for three times

the uncorrected value for the bulk, is shown in a comparison of the IR

transmission (Beckman IR-12 spectrophotometer) of the specimen after

Steps 1 and 2 (see Fig. 36). The absorption in the 3-fj.m region (curve 2

for Step 2) can be due to O-H and C-II. In view of the low pH employed,

such species of OH is not OH' as resulting from hydrolysis, but the

functional group of the polyhydric alcohol (glycerine). The C-O and

C-C vibrations would account for the absorption envelope at the 9. 5-|jLm

region which is seen to extend past 10. 6 fxm. The pertinent difference

121

TABLE XVIII

The Value of a in AJ (m = 0. 67 g, L i 0.27 cm) Versus Surface Treatment^'

Chronologi cal Order

Surface Treatment

Step 1

Step 2

Step 3

Step*4

Step 5

Step 6

After cleaving

Pv lished

Repoli shed

Repoli shed

Annealed

Polished

a, cm

0.0043

0.0067

0.013

0.0-5

0.0020

0.0015

The beam was positioned at the geometric center. Polishing was carried out with Li nde A i n ethanol .

T57(J

122

c OJ ••— E

• r—

et b 0)

OJ U T3 5^ C »^

■r- C0

+-> c:

2 '-- <C CM

CU O) OJ x: "o > co c 1-

•r- •!- 3 ^ —I o o —-

Q. J= ■•-> r—

^2 E 3;

•i- X3 CL U OJ 0) ^ T3 Q. in C

I/) -^ (D

r— O >— ro CL O

■M M lO C >, ^ ro ro

1- fO s 3

•*- <n ooioo

> c s- c O 3 O

•I- l_> •!- l/)^ -t-J

E o O i/i c 00 B ro ro JZ o S- -(-> CM

1— UJ X

•

n

O)

% NOISSIKISNVdl

123

"■■ NH

in the polishing procedures for Steps 1 and 2 resides in the volatility of

the alcohols: At 25 C, the vapor pressure of C^H^OH is 56 mm, while ,-4 2 5

C3H5^OH^3 is 4• ^ x 10 rnm. This result shows the importance of

avoiding incorporation in the surface (Beilby) layer of materials that

provide an M-O linkage (M = C or some other element); e. g, , the

abrasive (Al^C^) or even the wax coating of the lap. 25

We have shown that surface hydrolysis and hygroscopic behaviors

are negligible causes for surface absorption. However, the contribu-

tion of fabrication damage to the measurement of a is not negligible.

If such effect is not taken into consideration, the extrapolation of total

absorption to L = 0 can be misinterpreted as a contribution from the

surface. A contribution to surface absorption can come from the type

of chemicals employed in the polishing operation.

An improvement in the bulk absorption coefficient from

a = 1.6x10 cm" to Q = 3. 5 x 10" cm' was shown to result

from purification. The results of annealing and polishing studies indi-

cate that improved techniques in surface preparation are necessary to 3 1 achieve reproducibly a < 10" cm" .

124

■ ■—■"■■V-« I

APPENDIX. Ill

ABSORPTION SPECTROPHOTOMETRY IN THE CRYSTAL

The object is to derive a figure of merit for the method in terms 0f Cmin' the lower limit in the mole fraction of OH" in the crystal.

The optical absorption is described by,

1 - I0 • (43)

where L s thickness and o in cm" the absorption coefficient at the

peak wavelength In question and is given by,

ö Nc . (44)

Now, N in cm is the density of absorbers and 9 in cm is the absorp-

tion cross-section. The density of absorbers, N, is related to C, the

mole fraction of OH"(s),

/ N P \ N = (pÄTJ c («)

where the proportionality factor refers to the ion density of the host; N

0 = Avogadro number, p = crystal density, and [MX] = the molecular

or formula weight of the metal halide.

In absorption spectrophotometry, D, the experimentally mea-

sured optical density is given by

D - oL . (46)

125

It follows from eqs. (44) to (46) inclusive that,

C = [MX] • D

or (47)

As a figure of merit for the method we calculate C . . Real- min ?^

istic values are D . = 0. 01 and L = 1 cm. Now, N =6. 023 x 10 mm , o

For NaCl, [MX] = 58. 44 and p = 2. 165 g cm ; for KC1, [MX] = 74. 56 _3

and p 1.984 g cm . Therefore, from eq. (47),

C = 4. 48 x 10'25/cr for NaCl mm

C - 6. 24 x 10'25/ff for KC1. nun

126

- ■■- - - i —•■■ ---—— ■ ■ - ■ ■l.ll^^MH» -

wfw^m^^^^fimmm

A PPENDIX IV

SOLUTION ACIDIMETRY OF ALKALI IIALIDE CRYSTALS

The interpretation of acidimetric procedures considered below

is based on the assumption that when the alkali halide crystal is dis-

solved in water, the [OIl'ls)], expressed as a mole fraction, C, is

solely responsible for the shift in acidity.

Consider first the procedure based on a pH-shift measurement.

The fundamental constraint in the solvent (ILO) is.

[H+] [Oirj - K w

(48)

The ionization constant of the solvent, K , is assumed to be invariant

over the electrolyte concentration range employed, <1 M (M - molar).

Using the least hydrolyzed case as our C = 0; i. e. , the reference,

[H4] » 10-pHo and [Ohf] = 1 (T'^w-P1^). For thc hydroly/.ed case,

eq. (48) yields.

[10 •pHo - y] [lO-(pKw-PHo) t 10PC _ y] s 10-pKw ^ (49)

where y is the concentration reacted to attain equilibrium. As a mat-

ter of convenience, we have adopted a 1-M concentration of the halide

crystal so that C, the mole fraction of OH"(s) in the solid, is also the

molarity of OH in solution.

In the case of KC1, preliminary observations suggest a refer-

ence value of pH = i. 6 at 1 M in KC1. Since pK = 14. 0, it follows o w that.

l0-pHo >>10-(pKw-pHo) (50)

127

■ -

From eqs. (49) and (50),

-pllo 10-pC)

y ■ i ■ ' - i - 10-(pHotpC-0. 6021

(10-pHo H 10-pC)2

1/2

51

Note that the value of y is independent of pK . Calculations using w

eq. (51) with pH a 5.6 yield the working range in C:

Case 1. C- 10-5 yields y 1 0"5- 6028, pH - 7. H and pH-shift - - 2. 2.

Case 2. C 10"7 yields y- 10'7,00, pH - 5. 61 8 and pH-shift - t 0. 02.

-7 -5 Hence, the measurement is applicahlc to the range, 10 £ C ^ 10

using a pH meter with a precision of ±0. 01. Of course, this implies

that the measurement is carried out in a closed system which manifests

commensurate stability in the pH reading; i.e., negligible drift with

time.

Next, consider the use of an acid-base dye indicator, H.^,

where the ionization is given by.

c(l-o) CCH [H ] CO- 52)

where c is the molarity concentration of the dye, o is the degree of

ionization. and \H ll characterizes the hydrogen-ion concentration of i J 0

the electrolyte solution. The equilibrium constant of eq. (52) and the

total hydrogen-ion concentration, [H ], are related as follows.

K

IH'l "H^J" (53)

128

__ - - - - MBUMMMB mmm fl^Hfl^MH^^MI

11,111 ^•^mmm^mm^mt^^ i 11 i i i w^^^^^m*^^^^^^^^^^^

wliere,

I o

(a I [11] co • [H ]

(b) [<0'] co , and (54)

(O [H^ : cd-ol

The equilibrium constant can also b« expressed as follows,

K= 'TTi ■ T^T f'i I,, (55.

which is useful for determining K and fn'l , For co » firi Jo i JQ ' K K , where o

2 ,, co Ko rrr ' (56)

the intrinsic case; i. c. , the dye determines the hydrogen-ion

concentration.

The two hands of the optical ahsorption in solution are due

to H4 the species favored hy decreasing pH, and .9', the species

favored hy increasing pH. Note the following conservation in terms

of the concentrations shown in eqs. (54)(b) and (54)(c),

[HS] H [./-] - c . (57)

For simplicity, assume that the absorption bands are fairly resolved.

Assign to ILj? the extinction coefficient I. at peak wavelength absorp-

tion Xj and, similarly, to.Jf", € and \

129

«MM»Ma**

The absorbance of ^" is,

f, - «, Ll#-] (58)

where L is the cell thickness. A similar expression holds between

y-j and HÄ Since eqs. (53) and (57) yield,

K Ltl [Hh] c -[.*-]

(59)

it can be shown that.

dy2

d?H a 2-303 lzLc*^ (60)

Thus, the change in y with pH is most sensitive at o = 1/2.

Equation (60) also serves as a figure of merit for the method.

It was shown above that a pH-shift corresponding to dpH = 0. 01 accom- - -7 5 modates a working range in OH (s) of, 10 < C < 10' . A comparable

value of dy^ is easily obtained. If the choice of the dye is such that

o* 1/2, then dy^ ■ dpH implies according to eq. (60), t Lc < 2, and

from eqs. (54)(b) and (58), the absorbance is, y ■ 1. These operating

conditions are easily realized.

According to eq. (53), the measurement of [&~]/[H&], which

is also a /(I - a) by eqs. (54)(b) and (54)(c), yields [H+] in terms of K.

The measurement can be carried out using the two or just one of the

absorption bands. From eq. (58) and the analogous expression for

[H.?], it follows from eq. (53) that.

K

[H+] 1 - o (61

130

1_~_-a>____ ._^-_

'^•^mmmmmm^m^^m^^^^^ H^^K^^^HB

All the quantities at the right hand side of the equality are experimentally

determined. The one-absorption band method makes use of eqs. (58) and (59),

K

IH+] l2 Lc

(62;

Or, in terms of the Xj-band,

K

[ H+ ]

t Lc

yl

(63)

Determining 0 for various concentrations of the dye, c, for the

KC1 concentration employed (say. 1 M) yields K. From the K/[Hf]

value, [H+] is obtained and from eq. (54)(a), [u']0. The value of K

has to be determined only once by eq. (55) for I given dye and con-

centration of KC1. However, it can also be used to obtain [H ]o

directly, but this will involve a series of measurements with c as the

variable.

131

n i im ii

APPENDIX V

PLASTIC DEFORMATION OF IONIC FCC CRYSTAL

Given a face-center cubic (fee) ionic crystal, the problem is

to determine the orientation that will best resist plastic deformation.

Figure 57 illustrates the primary slip planes in such a

crystal at room temperature; the shading lines on each plane indicate

the slip direction in that plane. These slip systems are a Type (110)

<1 10> classification, of which the re are six. Burgers vectors b in

these slip systems will meet at angles of TT/3, rr/Z, or Zrr/^. Since

bj « b^ 4 bj for cos" (b, • b^/b^b^) > TT/2, and the energy of a dis-

location is proportional to b^ only one of these angles leads to asso-

ciation and locking of dislocations, viz. f/3.

One of the criteria for optimization of resistance to plastic

deformation is that cos ^ be minimum, i^ being the angle between the

direction of applied stress and the closest slip direction. The other

optimization criterion is that as many interferent slip systems as

possible be simultaneously active. The latter criterion implies that

tne direction of applied stress must be a principal axis of symmetry

of the crystal; this confines the selection to only low index directions,

ol which there are few, even in a cubic crystal.

Consider the two-fold, three-fold, and four-foH symmetric

directions. With the applied stress in the < 1 1 0> direction, although

one of the slip systems has a cos 6 of zero, another slip system has

a slip direction parallel with the direction of applied stress, and the

crystal is free to slip in that direction without interruption. Note in

this case that there are four other slip systems with a favorable cos cb

of one-half, and are interferent, but they assume a role secondary to

that of the < 1 10> slip direction.

With the applied stress in the < 1 11 > direction, three of the

slip systems have a cos * of zero, while the other three have a cos 6

of ^2/3. The latter three systems also are interferent.

132

■ ■ - •MaaMaMMMHll ■

With the applied stress in the <100> direction, two slip

systems have a cos * = 0, while the other four have cos <}) = \Jzll.

Among the latter four slip systems, however, each one is interferent

with only two of the other three and is cooperative with the remaining

slip system. This condition id similar to that where only two inter-

ferent slip systems exist.

From these considerations it may be concluded, therefore,

that the best direction for resistance to plastic deformation in an fee

ionic crystal is the <111> directioi. It should be pointed out, how-

ever, that such secondary factors as dislocation pinning by impurity

ions and reduction of dislocation mobility by the formation of jogs

have been disregarded; it is a possibility that.these factors may

predominate in real crystals.

1801 -2

s / y s ^Sfcfcw /

S s v^^^ W ÄpF- y / s /

Fig. 37. Slip Planes in an Ionic fee Crystal .

133

APPENDIX VI

CRYSTAL GROWTH AND EVALUATION OF METAL FLUORIDES

Because of their low solubility, only the refractory metal

fluorides were considered for use as windows in the 3- to 5-fjim

region, and only those materials which were potentially capable of

achieving an absorption coefficients 10' cm . Czochralski grcwth

from the melt was carried out under a reactive atmosphere processing

(RAP) of He + HF with a dewpoint parameter corresponding to a pres-

sure ratio of P^ Q/^HF = 10' •

Since no apparatus was available to measure calorimetrically

the absorption at 3 to 5 |j.m, the measurement of ß, the optical absorp-

tion coefficient at 10.6 |im, determined by adiabatic calorimetry

(App. II), was used as the index to the OH" content of the solid. This

index to anion purification through crystal growth was based on the

activity of the metal-to-oxygen vibration at the lO-mn region, which,

in turn, was a measure of the absorption at the 3-|im region arising

from the oxygen-to-hydrogen vibration.

Hydrolysis can readily occur at temperatures well below the

melting point, even in a refractory crystal Such as CaF?. A sig-

nificant lowering in mechanical strength is associated with hydrolysis,

a result which may be deduced in the comparison between LiF crystal

that is air-grown and that grown in vacuum. The latter growth

method yields a specimen that is less hydrolyzed, but still not com-

pletely free in OH" (Appendix I).

For convenience, we adopted the measurement of surface

microhardness (Knoop) as a qualitative indicator to the effect of RAP

on the mechanical properties. I! iwever, more direct evidence was

obtained after the contract ended.

The maximum strength reported in the literature for CaF^ is 8000 psi at Raytheon (Quarterly Report December 1970, ARPA Order 1180). Our RAP grown CaF^ gave a modulus of rupture of 12,000 psi.

134

The results of optical and mechanical evaluations are collected

in Table XIX. The standard deviations shown were derived from at

least four determinations. The crystals employed as starting materials

came from Ilarshaw (Cleveland, Ohio). Except in the case of SrF2,

the value of ß was obtained by means of the adiabatic calorimeter.

The Knoop values of PbF, show that the reported value of 200 kg/mm

in the 1967 Catalog of Harshaw Optical Crystals (Refer to Table 2,

page 1 i) is in error. Both ß and the Knoop hardness are useful indices to RAP.

Larger changes in the optical absorption could have resulted at the

^-lj.m region where the absorption cross section is larger. The results

show that the advantages of RAP are best realized if the procedure is

applied, right at the beginning; i.e. , the conversion of the oxide or

carbonate powder to the fluoride. This also points to the importance

of maintaining the specific surface of the material to reverse hydrolysis

prior to opeiating at the melting point (crystal growth). Consequently,

a one-step materials handling would be the preferred approach to RAP.

It is seen that, among the metal fluorides, BaF^ has the

lowest ß. Zb We considered the possibility that the tail of the absorp-

tion at 18 (xm contributed to ß."9 Consequently, we investigated two

approaches. C/.ochralski growth of BaF containing one mole percent BaCl2

was carried out. The resulting specimens have large scattering

losses (precipitate ?). Atmosphere control was a problem litte« the

correct ratio of HFlHCI to maintain over the melt was not known.

Crystal growth at a point of congruency was more desirable.

The phase diagram shows such a point, i.e., BaFCl.

Czochralski boules were grown from a molten equimolar mixture.

DTA characterization of the crystal with a DuPont Thermalyzer gave

a melting point of 1000 0C (under He) in good agreement with the

phase diagram. X-ray powder patterns showed the crystal to have

a tetragonal structure as reported; the lattice constants agreed with

the literature.

135

■ i ■ ■— ■—

c 0

fl) ^o o (VJ

fM ■H X £ Ov"" o — •« a 1

0 X +J x

£ M 00 ■H -H

If 00

*fH ^ rg O -H l-H f-H

1—1 Q0 ro ro fl) 0 (fl 1« iM O « * «U (0 M

■«-> -^ in * •> (Nl 2c 0] a 0 o c

-H -H M M

■H in PvJ fM

01 it u -H -H in ■H ■H

^-4

-H

in

u p f-H V

u IM c ic rvj fvj M M ^H oo »> » d t. UH

0 u 0

■ pen

dic

m

easu

2

f-H i E

ro O | 1

g 4J

(A

ü S i

m O

o • 'S 0 £

•4-» (0 £ d Tj

o 4-> | ^ o ^-v ■H C ti

la -H 1 TJ (M

0» v u u CO. r- \n o fM t^ U. B 00 fH r« (M rg ^H 0 U

1) • i—1

• o

• O

• o

• o « ^ Qt

ro o

X h o 0

k

.2 2 m 4j

< m

u IT)

oq 3 c 8 s ^H

ro 0 UJ ^H M d) B CT u O

03 i

£ P4 ■H 1| y <

0

_£ 00 X 1

-H in

• I l

0 c ' 0

(0

ro O

• o *

c o (0

Ml

■fi h 0)

a. 0 a

1

-H

o •H

■H

»—( •

I ■

in t

he

firs

t ro

w r

ef

e.

Th

e si

mil

ar

pai

the

sam

e m

ate

rial.

ß '»H

J3

■H

o ro

h ••-> n)

2 nl 1 c O ro o W . a c 0 u

1 00 c

At E

M <-H i—c 00 ro 0

<<H

0

O

II

f-H i • fH u ■*->

nl

03

H •

VH 0

C 0

c

0

VtH

o £ E

£ aT CH

I« 0) | g 1 4)

g 0 (Tt •

Tl C 0

c 0

0 T3 0) CO (0 "O -o _ o .. •IH

rtin

g m

; ed

uct

ion

C • H • fH

h 0

Pi Li

u >> 1« u u

0 0

of v

alu

es

hed

surf

a

surf

ace

o<

cn u 0) > c 0 o

o • fH to 0)

•fH

£ |

E V > c 0 u

m U m > B o u

<n ^H u in u 9 p E E

Met

al

Flu

ori

de

K h u

to (fl

m Th

e pai

th

e p

oli

cl

eaved

13

£ 0 ft

Fo

r th

e ch

emic

;

4->

£ 0 h

T3

£ 0 E

£ 0 ft

Ut ^^. -—^ i~ — ^-^, nl £ u s 'c? S

136

L - ■ - • - -

A polished crystal of BaFCl gave ß= 0.030 cm-1 which

degraded to 0. 110 cm" after an overnight exposure to the atmosphere.

The deterioration was not due to a surface layer change because after

repolishing, P = 0. 105 cm" . The microscope revealed the occurrence

of numerous cleavages in the bulk. The Knoop indentations were

poorly defined even at loads as low as 15 g. Edges of the indentations

crumbled badly and the hardness values covered a large spread, 120 to

210 kg/mm . In summary, the value of ß was reduced by a factor

of six from BaF, to BaFCl and although the material (BaFCl) was

also considerably harder, it was ultra brittle.

137

- ■ -

WWP^WIPPW"»"^* mPMWi««l MM. iiia^^MM« mimmmmmi^B^mmm

REFERENCES

1.

2.

3.

4.

5.

6.

7.

10.

11.

'Windows for High-Power Lasers,'' by F. Horrigan, C. Klein, R. Rudko, and D, Wilson, Microwaves, pp. 68-76 (January 1969).

"Strain Ageing in CdCl^-Doped Rock Salt,'' by L.M. Brown and P. L. Pratt, Phil. Mag. 8, 717 (1963).

"Impurity-Induced Infrared Absorption in Alkali Halide CO2 Laser Windows, ' by F.W. Patten, R. M. Garvey, and M. Hass, Mat. Res. Bull. 6, 1321 (1971).

"Der Einflusz von OH"-Ionen and Absorptions Spektrum und lonenleitfahigkeit von KCl-Einkristallen, " by B. Fritz, F. Luty, and F. Anger, Z. Phys. 174. 240 (1963).

"Influence of OH"lons on Infrared Absorption and Ionic Con- ductivity in Lithium Fluoride Crystals,'" by T.G. Stoebe, J. Phys. Chem. Solids 28. 1375(1967).

'U.V. and I. R. Absorption in OH"-Doped NaF, ' by M.L. Meistrich, J. Ph ^s. Chem. Solids 29, 1119(1968).

'"Distributer of Hydroxide Ions in Doped Alkali Halide Crystals, ' by T.G. Stoebe, J. Phys. Chem. Solids 31, 1291 (1970).

8. "Environmental Effects on the Mechanical Properties of Ionic Solids with Particular Reference to the Joffe Effect, "" by R.J. Stokes, T.L. Johnston, andC.H. Li, AIME Trans. 218. 655 (I960).

9. "Effect of Atmosphere Conditions on the Brittleness of NaCl,'1

byE.S. Machlin and G. T. Murray, J. App. Phys. 30 (No. 11). 1731 (1959). '

Influence of Room Trmperature Atmospheric Reaction Products on the Ductility of Sodium Chloride Crystals,'" by D. Otterson, J. Chem. Phys. 38 (No. 7), 1481 (1963).

Optical Properties of Alkali Halides Containing Hydroxyl Ions,'" by H.W. Et/.el and D.A. Patterson, Phys. Rev. 112 (No. 4), 1112 (1958).

138

12. "A Method for Purification and Growth of KC1 Single Crystals," byC.T. Butler, J.R. Russell, R.B. Quincy, Jr., and D.E. LaValle, Report No. ORNL-3906 (February 1966), Oak Ridge National Laboratory (Oak Ridge, Tennessee), operated by Union Carbide Corp., for the U.S. Atomic Energy Commis- sion, Contract No. W-7405-eng-26. See also references cited.

1 3. Preparation of Ultrapure Alkali Halide Single Crystals, " by F. Rosenberger, from Crystal Growth (Proceedings of 1966 ICCG Conference), edited by U.S. PeTser, see B20, p. 141 (Porgamon Press, 1967).

14. "Infrared and Ultraviolet OH Bands in Hydroxide-Doped KC1, KBr, NaCl and NaBr, " by T.I. Gie and M.V. Klein, Bull. Amer. Phys. Soc. Ser. II, 8, 230 (1963).

15. "Handbook of Analytical Chemistry," edited by L. Meites, First Edition, page 1-25 (McGraw-Hill 1963).

16. "On the Presence of NaOH in Crystalline NaCl, " by D. Otterson, J. Chem. Phys. ij_ (No. 1), 227 (I960).

17. "Growth of Large Sodium Chloride Crystals from Solution for Color-Center Studies, " by P.M. Gruzensky, J. Chem. Phys. 43 (No. 11), 5807(1965).

18. "Growth Poisoning of Lithium Fluoride from Aqueous Solutions. "

19. Cox, J Thomas, and Hass George (1964), "Antireflection coatings for optical and infrafed optical materials, "Physics of Thin Films 2 (Academic Press),

20. Macleod, H.A., (19b9) T'.nn Film Optical Filters, American Elsevier Publishing Company, Inc., New York.

2 1. Smith, S.O. and Seeley J. S. (1968) "Multilayer filters lor the region 0. 8 to 100 microns, "Final Scientific Report, Contract AF61(052)-833.

"« Pure NaCl powder is not hygroscopic. See H. Remy, Treatise on Inorganic Chemistry, Vol. I, (Elsevier Publish- ing Co. , 1956) p. 188. However, in the single crystal, surface damage can be effected with a very small weight change (<10-6 g/cm^).

23. Details of the materials preparation will be given in a separate publication.

24. R. Weil, J. Appl. Phys. 4j_(7), 5012 (1970).

139

mmmmmmmm

Z5. With vegetable waxes, the C-O linkage is present in the free fatty acids as well as in their esters with the higher alcohols.

26, W. Bontinck, "The Hydrolysis of Solid CaF/' Phvsica XXIV, 650 (1958). Z y

27.

28.

^•T^'TO5

-5

- 3allard' andK.A. McCarthy, "Mechanical

and Thorrnal Properties of Certain Optical Crystalline Materi- als, J. Opt. Soc 4l_ (No. 4), 215 (1951).

The value is in good agreement with the Raytheon determination reported in their Final Technical Report (Sept 1971)

ARPrOH^^o"/0/^1251' U-S- A"-yMi-üe Command, ARPA Order 1180 (cf. Fig. 15 on page 64).

29, W Kaiser W.G. Spitzer, R. H. Kaiser, and L.E. Howarth, Infrared Properties of CaF , SrF, and BaF,,"Phys. Rev. 127

(No. 6), 1950 (1962). 2 Z 2 '

30. E.M. Levin, C.R. Robbins, and II. F. McMurdie, "Phase Ma^?"13 ^.C^™8^," The American Ceramic Society (1964), see Fig. 1621. y

140

best available copy for all picturessigns and experimental results for coatings for alkali halides,...

Documents